Embodied intelligence refers to agents that sense，reason，act，and learn through a tight “body-environment-task” feedback loop. It is widely regarded as the most promising route to general-purpose artificial intelligence （AI）. Paradoxically，although model parameters，training data，and compute budgets have grown by four orders of magnitude in only five years，the evaluation ecosystem remains fragmented and largely irreproducible. The same functional capability （e.g.，“pick-and-place”） is formalized as object-detection mean average precision in one paper，as 3D intersection over union in a second，as task success rate in a third，and as human preference score in a fourth，with absolute gaps that exceed 10%—15% among “state-of-the-art” results. Environments，random seeds，physics parameters，sensor noise models，and success criteria are rarely open-sourced，and thus，the community cannot tell whether an apparent improvement originates from algorithmic innovation，data scale，evaluation cherry picking，or simple benchmark over-fitting. This uncertainty has become a critical bottleneck for scientific progress and industrial adoption. To address the gap，we conduct a systematic survey of 100+ papers published in IEEE/CVF Conference on Computer Vision and Pattern Recognition，IEEE International Conference on Robotics and Automation，Conference on Neural Information Processing Systems，Robotics： Science and Systems，International Conference on Learning Representations，International Journal of Robotic Research，and Journal of Machine Learning Research between 2020 and 2025. Then，we present the first holistic analysis of embodied AI evaluation. We organize the landscape into a three-tier pyramid： static datasets，simulation benchmarks，and real-robot trials. Each tier trades off cost，controllability，and ecological validity. Across all tiers，we decompose evaluation space into four tightly coupled stages： perception，cognition，decision，and execution. Then，we identify fine-grained capability dimensions along with 20+ quantitative metrics. For every dimension，we analyze the following： 1） the underlying methodological assumptions，2） the boundary conditions under which the metric is valid，3） the intrinsic limitations （e.g.，sim-to-real physics mismatch，dataset bias，human-label variance），and 4） the cost-credibility curve that determines when a researcher should ascend from static to simulation，and finally to the real world.All evaluations in this study are based on a theoretical “perception-cognition-decision-execution loop” framework. In embodied AI，perception is the process by which an agent converts raw，asynchronous，and often noisy signals from heterogeneous physical sensors，such as RGB-D cameras，LiDAR，tactile skins，joint encoders，force/torque，and inertial measurement unit units，into a registered，time-aligned digital twin that supplies metric 3D geometry and semantic labels. Cognition is then built on this torrent to distill structured，memory-enabled，and generalizable world models that encode object affordances，physical common sense，and relational and causal links，grounding free-form human language into the resulting latent space. Decision-making combines these world models with an explicit task goal and searches across symbolic，kinematic，and dynamic levels to output a coherent，risk-aware，and resource-efficient plan that is expressed as a hierarchy of subgoals，trajectories，and low-level set points. Finally，execution transforms this plan into real-world mechanical energy through high-rate feedback controllers that track desired motions or forces while compensating for actuator nonlinearities，shock，thermal limits，and human safety constraints，closing the perception-cognition-decision-execution loop that defines embodied AI. In contrast with humans，whose perception-cognition-decision-execution loop is seamlessly integrated，currently available embodied agents suffer structural flaws in every stage. Consequently，evaluation must decouple the four blocks and score accuracy，error，and latency separately to pinpoint bottlenecks. Static tests，i.e.，“written exams”，run on a GPU with a few megabytes of images or point clouds in minutes，but erase dynamics and physics. Simulation，i.e.，the “job interview”，adds interactive physics in tens of seconds but still drifts away from reality. Real robot trials，i.e.，“practical tests”，cost thousands，take days to deploy and minutes per run，but expose every sensing drift，prediction error，or control delay，offering the ultimate ground truth before field deployment. Therefore，by integrating the three-tier pyramid of static test-simulation-real robot evaluation into the closed perception-cognition-decision-execution loop，this work delivers the first systematic quantification of coverage，cost，and accuracy trade-offs across more than 100 embodied AI benchmarks，establishing a reproducible，traceable，and extensible evaluation baseline that not only exposes the exact stage at which state-of-the-art systems fail，but also provides a principled road map for allocating research budgets，standardizing protocols，and certifying safety before real-world deployment，marking a pivotal transition from empirical competition to scientific measurement，and accelerating the field’s development toward trustworthy，deployable，and general-purpose embodied intelligence. The detailed static-dynamic-real embodied intelligence evaluation can be accessed at https://opencompass.org.cn/embodied-intelligence.
具身智能通过“身体—环境—任务”闭环交互，被视为迈向通用人工智能的核心路径。然而，与模型参数和训练数据的高速扩张形成鲜明对照，其评测体系仍处于“任务碎片化、指标多元化、平台封闭化”的自发状态，造成同一功能在不同研究中的性能差异超过10%，却无法判定来源是算法创新、数据扩充还是评测偏差，严重阻碍了技术迭代与产业落地。本文围绕“从感知到执行”的完整链路，对2020—2025年间发表于CVPR（IEEE/CVF Conference on Computer Vision and Pattern Recognition）、ICRA（IEEE International Conference on Robotics and Automation）、NeurIPS（Conference on Neural Information Processing Systems）等顶级会议与 Nature、IJRR（International Journal of Robotic Research）、JMLR（Journal of Machine Learning Research）等期刊的百余篇文献进行系统梳理，首次提出“静态数据集—仿真平台—真实机器人”三级金字塔式评测范式，并从感知、认知、决策、执行4个环节拆解出20余项核心能力维度与量化指标，分别从方法学假设、适用边界、固有局限和成本—可信度曲线4个角度进行横向对比。本文进一步汇总了46套主流基在数据规模、任务类型、评估指标、开源程度和安全伦理考量等维度的差异。基于此，提出“能力导向、协议统一、三级协同”的未来框架：1）在任务层，由“功能对标”转向“能力对标”，建立可分解、可溯源、可加权的多维误差体系；2）在协议层，制定统一的场景描述、接口规范与指标定义，实现跨平台、跨任务、跨模型的可比性；3）在系统层，构建可远程接入、7 × 24 h运行的共享机器人集群，形成“仿真预训练—真机微调—在线更新”的闭环追踪，降低重复建设成本，打通从实验室到场景落地的“最后一公里”。最后讨论了安全伦理、文化偏见和能效评估等新维度如何纳入量化框架，并给出标准化路线图的短、中、长期目标。本文工作为具身智能从“技术涌现”走向“科学共识”提供了可操作的评测基础设施与参考范式，具体的静态—仿真—真机榜单展示在：https://opencompass.org.cn/embodied-intelligence。
, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1322960870430618526, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=1, ext=[AuthorCompanyExt(id=1322960870459978655, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870430618526, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Center of AI Evaluation，Shanghai AI Lab，Shanghai 200032，China), AuthorCompanyExt(id=1322960870476755872, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870430618526, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 上海人工智能实验室评测专项组，上海;200032)]), AuthorCompany(id=1322960870548059041, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=2, ext=[AuthorCompanyExt(id=1322960870560641954, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870548059041, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Information Science and Electronic Engineering，Shanghai Jiao Tong University，Shanghai 200240，China), AuthorCompanyExt(id=1322960870573224867, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870548059041, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 上海交通大学信息与电子工程学院，上海 200240)])])], keywords=[Keyword(id=1322960878617899999, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, orderNo=1, keyword=embodied intelligence), Keyword(id=1322960878680814560, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, orderNo=2, keyword=benchmark evaluation), Keyword(id=1322960878739534817, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, orderNo=3, keyword=perception-cognition-decision-execution), Keyword(id=1322960878810837986, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, orderNo=4, keyword=simulation), Keyword(id=1322960878877946851, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, orderNo=5, keyword=real-world validation), Keyword(id=1322960878957638628, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, orderNo=6, keyword=sim2real consistency), Keyword(id=1322960879028941797, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, orderNo=1, keyword=具身智能), Keyword(id=1322960879083467750, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, orderNo=2, keyword=评测体系), Keyword(id=1322960879175742439, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, orderNo=3, keyword=感知—认知—决策—执行), Keyword(id=1322960879255434216, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, orderNo=4, keyword=仿真平台), Keyword(id=1322960879335125993, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, orderNo=5, keyword=真机测试), Keyword(id=1322960879414817770, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, orderNo=6, keyword=仿真—现实一致性)], refs=[Reference(id=1322960883089027074, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2503.20020.pdf, language=null, rfNumber=null, rfOrder=0, authorNames=Abeyruwan S, Ainslie J, Alayrac J B, Arenas M G, Armstrong T, Balakrishna A, journalName=null, refType=null, unstructuredReference=AbeyruwanS，AinslieJ，AlayracJ B，ArenasM G，ArmstrongT，BalakrishnaA，et al. 2025. Gemini robotics： bringing AI into the physical world ［EB/OL］. ［2025-11-05］., articleTitle=Gemini robotics： bringing AI into the physical world, refAbstract=null), Reference(id=1322960883156135939, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=1, issue=1, pageStart=null, pageEnd=100008, url=null, language=null, rfNumber=null, rfOrder=1, authorNames=An Z L, Yu X Q, Wang C, Zhang J L, Li Y, Song C H, journalName=The Innovation Informatics, refType=null, unstructuredReference=AnZ L，YuX Q，WangC，ZhangJ L，LiY and SongC H. 2025. Embodied intelligence： recent advances and future perspectives. The Innovation Informatics，1（1）： #100008 ［DOI： 10.59717/j.xinn-inform.2025.100008］, articleTitle=Embodied intelligence： recent advances and future perspectives, refAbstract=null), Reference(id=1322960883223244804, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2003, volume=149, issue=1, pageStart=91, pageEnd=130, url=null, language=null, rfNumber=null, rfOrder=2, authorNames=Anderson M L, journalName=Artificial Intelligence, refType=null, unstructuredReference=AndersonM L. 2003. Embodied cognition： a field guide. Artificial Intelligence，149（1）： 91-130 ［DOI： 10.1016/S0004-3702（03）00054-7］, articleTitle=Embodied cognition： a field guide, refAbstract=null), Reference(id=1322960883281965061, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=19107, pageEnd=19117, url=null, language=null, rfNumber=null, rfOrder=3, authorNames=Azuma D, Miyanishi T, Kurita S, Kawanabe M, journalName=null, refType=null, unstructuredReference=AzumaD，MiyanishiT，KuritaS and KawanabeM. 2022. Scan Q A： 3D question answering for spatial scene understanding//Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans，USA： IEEE： 19107-19117, articleTitle=Scan Q A： 3D question answering for spatial scene understanding, refAbstract=null), Reference(id=1322960883349073926, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2503.15558.pdf, language=null, rfNumber=null, rfOrder=4, authorNames=Azzolini A, Bai J J, Brandon H, Cao J X, Chattopadhyay P, Chen H Y, journalName=null, refType=null, unstructuredReference=AzzoliniA，BaiJ J，BrandonH，CaoJ X，ChattopadhyayP，ChenH Y，et al. 2025. Cosmos-Reason1： from physical common sense to embodied reasoning ［EB/OL］. ［2025-11-05］., articleTitle=Cosmos-Reason1： from physical common sense to embodied reasoning, refAbstract=null), Reference(id=1322960883424571399, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=11288, pageEnd=11309, url=null, language=null, rfNumber=null, rfOrder=5, authorNames=Ballout M, Wilfred O, Yaghoubi S, Abdelmoneim N M, Mayer J, Bruni E, journalName=Findings of the Association for Computational Linguistics： EMNLP 2025, refType=null, unstructuredReference=BalloutM，WilfredO，YaghoubiS，AbdelmoneimN M，MayerJ and BruniE. 2025. Can you SPLICE it together？ A human curated benchmark for probing visual reasoning in VLMs//Findings of the Association for Computational Linguistics： EMNLP 2025. Suzhou，China： Association for Computational Linguistics： 11288-11309 ［DOI： 10.18653/v1/2025.findings-emnlp.604］, articleTitle=Can you SPLICE it together？ A human curated benchmark for probing visual reasoning in VLMs, refAbstract=null), Reference(id=1322960883504263176, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=6, authorNames=Belkhale S, Ding T L, Xiao T, Sermanet P, Vuong Q, Tompson J, journalName=Robotics： Science and Systems XX, refType=null, unstructuredReference=BelkhaleS，DingT L，XiaoT，SermanetP，VuongQ，TompsonJ，et al. 2024. RT-H： action hierarchies using language//Robotics： Science and Systems XX. Delft，the Netherlands： ［s.n.］, articleTitle=RT-H： action hierarchies using language, refAbstract=null), Reference(id=1322960883579760649, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=7, authorNames=Brohan A, Brown N, Carbajal J, Chebotar Y, Dabis J, Finn C, journalName=Robotics： Science and Systems XIX, refType=null, unstructuredReference=BrohanA，BrownN，CarbajalJ，ChebotarY，DabisJ，FinnC，et al. 2023. RT-1： robotics transformer for real-world control at scale//Robotics： Science and Systems XIX. Daegu，Korea（South）： ［s.n.］, articleTitle=RT-1： robotics transformer for real-world control at scale, refAbstract=null), Reference(id=1322960883672035338, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.17321.pdf, language=null, rfNumber=null, rfOrder=8, authorNames=Budzianowski P, Wiśnios E, Tyrolski M, Góral G, Kulakov I, Petrenko V, journalName=null, refType=null, unstructuredReference=BudzianowskiP，WiśniosE，TyrolskiM，GóralG，KulakovI，PetrenkoV，et al. 2025. OpenGVL-benchmarking visual temporal progress for data curation ［EB/OL］. ［2025-11-05］., articleTitle=OpenGVL-benchmarking visual temporal progress for data curation, refAbstract=null), Reference(id=1322960883751727115, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.22737.pdf, language=null, rfNumber=null, rfOrder=9, authorNames=Cai J, Yang K N, Fu L, Ding J M, Li J L, Sun H M, journalName=null, refType=null, unstructuredReference=CaiJ，YangK N，FuL，DingJ M，LiJ L，SunH M，et al. 2025. CompareBench： a benchmark for visual comparison reasoning in vision-language models ［EB/OL］. ［2025-11-05］., articleTitle=CompareBench： a benchmark for visual comparison reasoning in vision-language models, refAbstract=null), Reference(id=1322960883814641676, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2017, volume=null, issue=null, pageStart=667, pageEnd=676, url=null, language=null, rfNumber=null, rfOrder=10, authorNames=Chang A, Dai A, Funkhouser T, Halber M, Niessner M, Savva M, journalName=null, refType=null, unstructuredReference=ChangA，DaiA，FunkhouserT，HalberM，NiessnerM，SavvaM，et al. 2017. Matterport3D： learning from RGB-D data in indoor environments//Proceedings of 2017 International Conference on 3D Vision （3DV）. Qingdao，China： IEEE： 667-676 ［DOI： 10.1109/3DV.2017.00081］, articleTitle=Matterport3D： learning from RGB-D data in indoor environments, refAbstract=null), Reference(id=1322960883881750541, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=null, pageStart=202, pageEnd=221, url=null, language=null, rfNumber=null, rfOrder=11, authorNames=Chen D Z, Chang A X, Nießner M, journalName=null, refType=null, unstructuredReference=ChenD Z，ChangA X and NießnerM. 2020. ScanRefer： 3D object localization in RGB-D scans using natural language//Proceedings of the 16th European Conference on Computer Vision — ECCV 2020. Glasgow，UK： Springer： 202-221 ［DOI： 10.1007/978-3-030-58565-5_13］, articleTitle=ScanRefer： 3D object localization in RGB-D scans using natural language, refAbstract=null), Reference(id=1322960883948859406, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=3192, pageEnd=3202, url=null, language=null, rfNumber=null, rfOrder=12, authorNames=Chen D Z, Gholami A, Nießner M and Chang A X, journalName=null, refType=null, unstructuredReference=ChenD Z，GholamiA，Nießner M and ChangA X. 2021. Scan2Cap： context-aware dense captioning in RGB-D scans//Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville，USA： IEEE： 3192-3202 ［DOI： 10.1109/CVPR46437.2021.00321］, articleTitle=Scan2Cap： context-aware dense captioning in RGB-D scans, refAbstract=null), Reference(id=1322960884020162575, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025a, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.25885.pdf, language=null, rfNumber=null, rfOrder=13, authorNames=Chen R L, Sun Y Q, Wang J H, Lyu M Y, Zhang Q, Zeng Y, journalName=null, refType=null, unstructuredReference=ChenR L，SunY Q，WangJ H，LyuM Y，ZhangQ and ZengY. 2025a. SafeMind： benchmarking and mitigating safety risks in embodied LLM agents ［EB/OL］. ［2025-11-05］., articleTitle=SafeMind： benchmarking and mitigating safety risks in embodied LLM agents, refAbstract=null), Reference(id=1322960884091465744, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025b, volume=0, issue=null, pageStart=2, pageEnd=null, url=https://arxiv.org/pdf/2506.18088.pdf, language=null, rfNumber=null, rfOrder=14, authorNames=Chen T X, Chen Z X, Chen B J, Cai Z J, Liu Y B, Li Z X, journalName=null, refType=null, unstructuredReference=ChenT X，ChenZ X，ChenB J，CaiZ J，LiuY B，LiZ X，et al. 2025b. RoboTwin 2.0： a scalable data generator and benchmark with strong domain randomization for robust bimanual robotic manipulation ［EB/OL］. ［2025-11-05］., articleTitle=RoboTwin, refAbstract=null), Reference(id=1322960885676912657, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025c, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.20148.pdf, language=null, rfNumber=null, rfOrder=15, authorNames=Chen Z, Gholami A, Nießner M and Chang A X, journalName=null, refType=null, unstructuredReference=ChenZ，GholamiA，Nießner M and ChangA X. 2025c. MineAnyBuild： benchmarking spatial planning for open-world ai agents［EB/OL］.［2025-11-05］., articleTitle=MineAnyBuild： benchmarking spatial planning for open-world ai agents, refAbstract=null), Reference(id=1322960885769187346, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2508.08706.pdf, language=null, rfNumber=null, rfOrder=16, authorNames=Cheng Z X, Zhang Y Q, Zhang W K, Li H Y, Wang K Y, Song L, journalName=null, refType=null, unstructuredReference=ChengZ X，ZhangY Q，ZhangW K，LiH Y，WangK Y，SongL，et al. 2025. OmniVTLA： vision-tactile-language-action model with semantic-aligned tactile sensing ［EB/OL］. ［2025-11-05］., articleTitle=OmniVTLA： vision-tactile-language-action model with semantic-aligned tactile sensing, refAbstract=null), Reference(id=1322960885848879123, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2409.14277.pdf, language=null, rfNumber=null, rfOrder=17, authorNames=Chia Y K, Sun Q, Bing L D, Poria S, journalName=null, refType=null, unstructuredReference=ChiaY K，SunQ，BingL D and PoriaS. 2024. Can-Do！ A dataset and neuro-symbolic grounded framework for embodied planning with large multimodal models ［EB/OL］. ［2025-11-05］., articleTitle=Can-Do！ A dataset and neuro-symbolic grounded framework for embodied planning with large multimodal models, refAbstract=null), Reference(id=1322960885903405076, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.03233.pdf, language=null, rfNumber=null, rfOrder=18, authorNames=Deng S L, Yan M, Wei S L, Ma H X, Yang Y X, Chen J Y, journalName=null, refType=null, unstructuredReference=DengS L，YanM，WeiS L，MaH X，YangY X，ChenJ Y，et al. 2025. GraspVLA： a grasping foundation model pre-trained on billion-scale synthetic action data ［EB/OL］. ［2025-11-05］., articleTitle=GraspVLA： a grasping foundation model pre-trained on billion-scale synthetic action data, refAbstract=null), Reference(id=1322960885966319637, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=7147, pageEnd=7153, url=null, language=null, rfNumber=null, rfOrder=19, authorNames=Du Y W, Feng S Y, Cort C G, Vela P A, journalName=null, refType=null, unstructuredReference=DuY W，FengS Y，CortC G and VelaP A. 2025. Task-driven SLAM benchmarking for robot navigation//Proceedings of 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Hangzhou，China： IEEE： 7147-7153 ［DOI： 10.1109/IROS60139.2025.11246415］, articleTitle=Task-driven SLAM benchmarking for robot navigation, refAbstract=null), Reference(id=1322960886033428502, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2022, volume=6, issue=2, pageStart=230, pageEnd=244, url=null, language=null, rfNumber=null, rfOrder=20, authorNames=Duan J F, Yu S, Tan H L, Zhu H Y, Tan C, journalName=IEEE Transactions on Emerging Topics in Computational Intelligence, refType=null, unstructuredReference=DuanJ F，YuS，TanH L，ZhuH Y and TanC. 2022. A survey of embodied AI： from simulators to research tasks. IEEE Transactions on Emerging Topics in Computational Intelligence，6（2）： 230-244 ［DOI： 10.1109/TETCI.2022.3141105］, articleTitle=A survey of embodied AI： from simulators to research tasks, refAbstract=null), Reference(id=1322960886092148759, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=653, pageEnd=660, url=null, language=null, rfNumber=null, rfOrder=21, authorNames=Fang H S, Fang H J, Tang Z Y, Liu J R, Wang C X, Wang J B, journalName=null, refType=null, unstructuredReference=FangH S，FangH J，TangZ Y，LiuJ R，WangC X，WangJ B，et al. 2024. RH20T： a comprehensive robotic dataset for learning diverse skills in one-shot//Proceedings of 2024 IEEE International Conference on Robotics and Automation （ICRA）. Yokohama，Japan： IEEE： 653-660 ［DOI： 10.1109/ICRA57147.2024.10611615］, articleTitle=RH20T： a comprehensive robotic dataset for learning diverse skills in one-shot, refAbstract=null), Reference(id=1322960886159257624, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://action-net.org/, language=null, rfNumber=null, rfOrder=22, authorNames=Fourier ActionNet Team and Mu Y, journalName=null, refType=null, unstructuredReference=Fourier ActionNet Team and Mu Y. 2025. ActionNet： a dataset for dexterous bimanual manipulation ［EB/OL］. ［2025-11-05］., articleTitle=ActionNet： a dataset for dexterous bimanual manipulation, refAbstract=null), Reference(id=1322960886217977881, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.23690.pdf, language=null, rfNumber=null, rfOrder=23, authorNames=Gao S Y, Yao J S, Wen H Y, Guo Y H, Liu Z M, Huang H Y, journalName=null, refType=null, unstructuredReference=GaoS Y，YaoJ S，WenH Y，GuoY H，LiuZ M and HuangH Y. 2025. HomeSafeBench： a benchmark for embodied vision-language models in free-exploration home safety inspection［EB/OL］. ［2025-11-05］., articleTitle=HomeSafeBench： a benchmark for embodied vision-language models in free-exploration home safety inspection, refAbstract=null), Reference(id=1322960886285086746, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2012, volume=null, issue=null, pageStart=3354, pageEnd=3361, url=null, language=null, rfNumber=null, rfOrder=24, authorNames=Geiger A, Lenz P, Urtasun R, journalName=null, refType=null, unstructuredReference=GeigerA，LenzP and UrtasunR. 2012. Are we ready for autonomous driving？ The KITTI vision benchmark suite//Proceedings of 2012 IEEE Conference on Computer Vision and Pattern Recognition. Providence，USA： IEEE： 3354-3361 ［DOI： 10.1109/CVPR.2012.6248074］, articleTitle=Are we ready for autonomous driving？ The KITTI vision benchmark suite, refAbstract=null), Reference(id=1322960886348001307, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2504.18904.pdf, language=null, rfNumber=null, rfOrder=25, authorNames=Geng H R, Wang F S, Wei S L, Li Y Y, Wang B J, An B S, journalName=null, refType=null, unstructuredReference=GengH R，WangF S，WeiS L，LiY Y，WangB J，AnB S，et al. 2025. RoboVerse： towards a unified platform，dataset and benchmark for scalable and generalizable robot learning ［EB/OL］. ［2025-11-05］., articleTitle=RoboVerse： towards a unified platform，dataset and benchmark for scalable and generalizable robot learning, refAbstract=null), Reference(id=1322960886406721564, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2510.07869.pdf, language=null, rfNumber=null, rfOrder=26, authorNames=Gu J W, Wu Z H, Si P X, Qiu S, Feng Y K, Sun L Y, journalName=null, refType=null, unstructuredReference=GuJ W，WuZ H，SiP X，QiuS，FengY K，SunL Y，et al. 2025. USIM and U0： a vision-language-action dataset and model for general underwater robots ［EB/OL］. ［2025-11-05］., articleTitle=USIM and U0： a vision-language-action dataset and model for general underwater robots, refAbstract=null), Reference(id=1322960886469636125, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=27, authorNames=Gu J Y, Kirmani S, Wohlhart P, Lu Y, Gonzalez Arenas M, Rao K, journalName=null, refType=null, unstructuredReference=GuJ Y，KirmaniS，WohlhartP，LuY，Gonzalez ArenasM，RaoK，et al. 2024. RT-Trajectory： robotic task generalization via hindsight trajectory sketches//Proceedings of the 12th International Conference on Learning Representations. Vienna，Austria： OpenReview.net, articleTitle=RT-Trajectory： robotic task generalization via hindsight trajectory sketches, refAbstract=null), Reference(id=1322960886532550686, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=2, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=28, authorNames=Gu J Y, Xiang F B, Li X L, Ling Z, Liu X Q, Mu T Z, journalName=ManiSkill, refType=null, unstructuredReference=GuJ Y，XiangF B，LiX L，LingZ，LiuX Q，MuT Z，et al. 2023. ManiSkill2： a unified benchmark for generalizable manipulation skills//Proceedings of the 11th International Conference on Learning Representations. Kigali，Rwanda： OpenReview.net, articleTitle=a unified benchmark for generalizable manipulation skills, refAbstract=null), Reference(id=1322960886595465247, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.18229.pdf, language=null, rfNumber=null, rfOrder=29, authorNames=Guo M N, Wu M W, He J R, Li S X, Li H F, Tao C, journalName=null, refType=null, unstructuredReference=GuoM N，WuM W，HeJ R，LiS X，LiH F and TaoC. 2025. BEDI： a comprehensive benchmark for evaluating embodied agents on UAVs［EB/OL］. ［2025-11-05］., articleTitle=BEDI： a comprehensive benchmark for evaluating embodied agents on UAVs, refAbstract=null), Reference(id=1322960886654185504, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2506.06677.pdf, language=null, rfNumber=null, rfOrder=30, authorNames=Han S H, Qiu B X, Liao Y, Huang S Y, Gao C, Yan S C, journalName=null, refType=null, unstructuredReference=HanS H，QiuB X，LiaoY，HuangS Y，GaoC，YanS C，et al. 2025. RoboCerebra： a large-scale benchmark for long-horizon robotic manipulation evaluation ［EB/OL］. ［2025-11-05］., articleTitle=RoboCerebra： a large-scale benchmark for long-horizon robotic manipulation evaluation, refAbstract=null), Reference(id=1322960886712905761, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2504.14650.pdf, language=null, rfNumber=null, rfOrder=31, authorNames=Huang Y T, Ding L L, Tang Z P, Wang T F, Lin X R, Zhang W Y, journalName=null, refType=null, unstructuredReference=HuangY T，DingL L，TangZ P，WangT F，LinX R，ZhangW Y，et al. 2025. A framework for benchmarking and aligning task-planning safety in LLM-based embodied agents ［EB/OL］. ［2025-11-05］., articleTitle=A framework for benchmarking and aligning task-planning safety in LLM-based embodied agents, refAbstract=null), Reference(id=1322960886780014626, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://internrobotics.shlab.org.cn/opendataset.html, language=null, rfNumber=null, rfOrder=32, authorNames=Intern Robotics Team, journalName=null, refType=null, unstructuredReference=Intern Robotics Team. 2025. InternData： a large-scale multimodal synthetic and real hybrid dataset for embodied intelligence［EB/OL］. ［2025-11-05］., articleTitle=InternData： a large-scale multimodal synthetic and real hybrid dataset for embodied intelligence, refAbstract=null), Reference(id=1322960886842929187, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2020, volume=5, issue=2, pageStart=3019, pageEnd=3026, url=null, language=null, rfNumber=null, rfOrder=33, authorNames=James S, Ma Z C, Arrojo D R, Davison A J, journalName=IEEE Robotics and Automation Letters, refType=null, unstructuredReference=JamesS，MaZ C，ArrojoD R and DavisonA J. 2020. RLBench： the robot learning benchmark and learning environment. IEEE Robotics and Automation Letters，5（2）： 3019-3026, articleTitle=RLBench： the robot learning benchmark and learning environment, refAbstract=null), Reference(id=1322960886914232356, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2503.11117.pdf, language=null, rfNumber=null, rfOrder=34, authorNames=Jiang K X, Liu Y, Chen W X, Luo J Z, Chen Z L, Pan L, journalName=null, refType=null, unstructuredReference=JiangK X，LiuY，ChenW X，LuoJ Z，ChenZ L，PanL，et al. 2025. Beyond the destination： a novel benchmark for exploration-aware embodied question answering ［EB/OL］. ［2025-11-05］., articleTitle=Beyond the destination： a novel benchmark for exploration-aware embodied question answering, refAbstract=null), Reference(id=1322960886981341221, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=1996, volume=4, issue=1, pageStart=237, pageEnd=285, url=null, language=null, rfNumber=null, rfOrder=35, authorNames=Kaelbling L P, Littman M L, Moore A W, journalName=Journal of Artificial Intelligence Research, refType=null, unstructuredReference=KaelblingL P，LittmanM L and MooreA W. 1996. Reinforcement learning： a survey. Journal of Artificial Intelligence Research，4（1）： 237-285, articleTitle=Reinforcement learning： a survey, refAbstract=null), Reference(id=1322960887044255782, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=11257, pageEnd=11264, url=null, language=null, rfNumber=null, rfOrder=36, authorNames=Kästner L, Carstens R, Zeng H J, Kmiecik J, Bhuiyan T, Khorsandhi N, journalName=null, refType=null, unstructuredReference=KästnerL，CarstensR，ZengH J，KmiecikJ，BhuiyanT，KhorsandhiN，et al. 2023. Arena-Rosnav 2.0： a development and benchmarking platform for robot navigation in highly dynamic environments//Proceedings of 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Detroit，USA： IEEE： 11257-11264 ［DOI： 10.1109/IROS55552.2023.10342152］, articleTitle=Arena-Rosnav 2.0： a development and benchmarking platform for robot navigation in highly dynamic environments, refAbstract=null), Reference(id=1322960887107170343, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=16373, pageEnd=16383, url=null, language=null, rfNumber=null, rfOrder=37, authorNames=Khanna M, Ramrakhya R, Chhablani G, Yenamandra S, Gervet T, Chang M, journalName=null, refType=null, unstructuredReference=KhannaM，RamrakhyaR，ChhablaniG，YenamandraS，GervetT，ChangM，et al. 2024. GOAT-Bench： a benchmark for multi-modal lifelong navigation//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle，USA： IEEE： 16373-16383 ［DOI： 10.1109/CVPR52733.2024.01549］, articleTitle=GOAT-Bench： a benchmark for multi-modal lifelong navigation, refAbstract=null), Reference(id=1322960887182667816, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=38, authorNames=Khazatsky A, Pertsch K, Nair S, Balakrishna A, Dasari S, Karamcheti S, journalName=Robotics： Science and Systems XX, refType=null, unstructuredReference=KhazatskyA，PertschK，NairS，BalakrishnaA，DasariS，KaramchetiS，et al. 2024. DROID： a large-scale in-the-wild robot manipulation dataset//Robotics： Science and Systems XX. Delft，the Netherlands： ［s.n.］, articleTitle=DROID： a large-scale in-the-wild robot manipulation dataset, refAbstract=null), Reference(id=1322960887262359593, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2017, volume=40, issue=null, pageStart=null, pageEnd=253, url=null, language=null, rfNumber=null, rfOrder=39, authorNames=Lake B M, Ullman T D, Tenenbaum J B, Gershman S J, journalName=Behavioral and Brain Sciences, refType=null, unstructuredReference=LakeB M，UllmanT D，TenenbaumJ B and GershmanS J. 2017. Building machines that learn and think like people. Behavioral and Brain Sciences，40： #253 ［DOI： 10.1017/S0140525X16001837］, articleTitle=Building machines that learn and think like people, refAbstract=null), Reference(id=1322960887329468458, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2015, volume=521, issue=7553, pageStart=436, pageEnd=444, url=null, language=null, rfNumber=null, rfOrder=40, authorNames=LeCun Y, Bengio Y, Hinton G, journalName=Nature, refType=null, unstructuredReference=LeCunY，BengioY and HintonG. 2015. Deep learning. Nature，521（7553）： 436-444 ［DOI： 10.1038/nature14539］, articleTitle=Deep learning, refAbstract=null), Reference(id=1322960887413354539, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=5912, pageEnd=5922, url=null, language=null, rfNumber=null, rfOrder=41, authorNames=Lee J, Miyanishi T, Kurita S, Sakamoto K, Azuma D, Matsuo Y, journalName=CityNav： a large-scale dataset for real-world aerial navigation/, refType=null, unstructuredReference=LeeJ，MiyanishiT，KuritaS，SakamotoK，AzumaD，MatsuoY，et al. 2025. CityNav： a large-scale dataset for real-world aerial navigation//Proceedings of 2025 IEEE/CVF International Conference on Computer Vision. 5912-5922, articleTitle=2025, refAbstract=null), Reference(id=1322960887480463404, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025a, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.16815.pdf, language=null, rfNumber=null, rfOrder=42, authorNames=Li C Y, Xiao J H, Zhang J B, Wen F R, Zhang Z C, Tian Y, journalName=null, refType=null, unstructuredReference=LiC Y，XiaoJ H，ZhangJ B，WenF R，ZhangZ C，TianY，et al. 2025a. Image quality assessment for embodied AI ［EB/OL］. ［2025-11-05］., articleTitle=Image quality assessment for embodied AI, refAbstract=null), Reference(id=1322960887564349485, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025b, volume=19, issue=7, pageStart=1349, pageEnd=1361, url=null, language=null, rfNumber=null, rfOrder=43, authorNames=Li C Y, Zhang J B, Zhang Z C, Wu H N, Tian Y, Sun W, journalName=IEEE Journal of Selected Topics in Signal Processing, refType=null, unstructuredReference=LiC Y，ZhangJ B，ZhangZ C，WuH N，TianY，SunW，et al. 2025b. R-Bench： are your large multimodal model robust to real-world corruptions？ IEEE Journal of Selected Topics in Signal Processing，19（7）： 1349-1361 ［DOI： 10.1109/JSTSP.2025.3558652］, articleTitle=R-Bench： are your large multimodal model robust to real-world corruptions？, refAbstract=null), Reference(id=1322960887648235566, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025c, volume=null, issue=null, pageStart=3705, pageEnd=3728, url=null, language=null, rfNumber=null, rfOrder=44, authorNames=Li X, Hsu K, Gu J, Pertsch K, Mees O, Walke H R, journalName=null, refType=null, unstructuredReference=LiX，HsuK，GuJ，PertschK，MeesO，WalkeH R，et al. 2025c. Evaluating real-world robot manipulation policies in simulation//Proceedings of the 8th Conference on Robot Learning. Munich，Germany： PMLR： 3705-3728, articleTitle=Evaluating real-world robot manipulation policies in simulation, refAbstract=null), Reference(id=1322960887723733039, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2508.10399.pdf, language=null, rfNumber=null, rfOrder=45, authorNames=Liang W L, Zhou R, Ma Y, Zhang B, Li S L, Liao Y J, journalName=null, refType=null, unstructuredReference=LiangW L，ZhouR，MaY，ZhangB，LiS L，LiaoY J，et al. 2025. Large model empowered embodied AI： a survey on decision-making and embodied learning ［EB/OL］. ［2025-11-05］., articleTitle=Large model empowered embodied AI： a survey on decision-making and embodied learning, refAbstract=null), Reference(id=1322960887795036208, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=#1939, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=46, authorNames=Liu B, Zhu Y F, Gao C K, Feng Y H, Liu Q, Zhu Y K, journalName=null, refType=null, unstructuredReference=LiuB，ZhuY F，GaoC K，FengY H，LiuQ，ZhuY K，et al. 2023. LIBERO： benchmarking knowledge transfer for lifelong robot learning//Proceedings of the 37th International Conference on Neural Information Processing Systems. New Orleans，USA： Curran Associates Inc.： #1939, articleTitle=LIBERO： benchmarking knowledge transfer for lifelong robot learning, refAbstract=null), Reference(id=1322960887862145073, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025a, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.22356.pdf, language=null, rfNumber=null, rfOrder=47, authorNames=Liu E G, Liang S Y, Lu L M, Zeng X Y, Cao X C, Liu A S, journalName=null, refType=null, unstructuredReference=LiuE G，LiangS Y，LuL M，ZengX Y，CaoX C，LiuA S，et al. 2025a. RoboView-Bias： benchmarking visual bias in embodied agents for robotic manipulation ［EB/OL］. ［2025-11-05］., articleTitle=RoboView-Bias： benchmarking visual bias in embodied agents for robotic manipulation, refAbstract=null), Reference(id=1322960887937642546, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025b, volume=null, issue=null, pageStart=10607, pageEnd=10615, url=null, language=null, rfNumber=null, rfOrder=48, authorNames=Liu H R, Wan W K, Yu X Q, Li M H, Zhang J Z, Zhao B, journalName=null, refType=null, unstructuredReference=LiuH R，WanW K，YuX Q，LiM H，ZhangJ Z，ZhaoB，et al. 2025b. Na Vid-4D： unleashing spatial intelligence in egocentric RGB-D videos for vision-and-language navigation//Proceedings of 2025 IEEE International Conference on Robotics and Automation. Atlanta，USA： IEEE： 10607-10615 ［DOI： 10.1109/ICRA55743.2025.11128467］, articleTitle=Na Vid-4D： unleashing spatial intelligence in egocentric RGB-D videos for vision-and-language navigation, refAbstract=null), Reference(id=1322960888013140019, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025c, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2510.00405.pdf, language=null, rfNumber=null, rfOrder=49, authorNames=Liu J Y, Zhou J M, Ye K, Lin K Y, Wang A, Liang J W, journalName=null, refType=null, unstructuredReference=LiuJ Y，ZhouJ M，YeK，LinK Y，WangA and LiangJ W. 2025c. EgoTraj-Bench： towards robust trajectory prediction under ego-view noisy observations ［EB/OL］. ［2025-11-05］., articleTitle=EgoTraj-Bench： towards robust trajectory prediction under ego-view noisy observations, refAbstract=null), Reference(id=1322960888084443188, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2506.16402.pdf, language=null, rfNumber=null, rfOrder=50, authorNames=Lu X Y, Chen Z, Hu X H, Zhou Y J, Zhang W C, Liu D R, journalName=null, refType=null, unstructuredReference=LuX Y，ChenZ，HuX H，ZhouY J，ZhangW C，LiuD R，et al. 2025. IS-Bench： evaluating interactive safety of VLM-driven embodied agents in daily household tasks ［EB/OL］. ［2025-11-05］., articleTitle=IS-Bench： evaluating interactive safety of VLM-driven embodied agents in daily household tasks, refAbstract=null), Reference(id=1322960888155746357, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=44, issue=4, pageStart=592, pageEnd=606, url=null, language=null, rfNumber=null, rfOrder=51, authorNames=Luo J L, Xu C, Liu F C, Tan L M, Lin Z P, Wu J, journalName=The International Journal of Robotics Research, refType=null, unstructuredReference=LuoJ L，XuC，LiuF C，TanL M，LinZ P，WuJ，et al. 2025. FMB： a functional manipulation benchmark for generalizable robotic learning. The International Journal of Robotics Research，44（4）： 592-606 ［DOI： 10.1177/02783649241276017］, articleTitle=FMB： a functional manipulation benchmark for generalizable robotic learning, refAbstract=null), Reference(id=1322960888222855222, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2506.08708.pdf, language=null, rfNumber=null, rfOrder=52, authorNames=Ma L, Wen J J, Lin M, Xu R T, Liang X W, Lin B Q, journalName=null, refType=null, unstructuredReference=MaL，WenJ J，LinM，XuR T，LiangX W，LinB Q，et al. 2025. PhyBlock： a progressive benchmark for physical understanding and planning via 3D block assembly ［EB/OL］. ［2025-11-05］., articleTitle=PhyBlock： a progressive benchmark for physical understanding and planning via 3D block assembly, refAbstract=null), Reference(id=1322960888298352695, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=3, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=53, authorNames=Ma X J, Yong S L, Zheng Z L, Li Q, Liang Y T, Zhu S C, journalName=SQA, refType=null, unstructuredReference=MaX J，YongS L，ZhengZ L，LiQ，LiangY T，ZhuS C，et al. 2023. SQA3D： situated question answering in 3D scenes//Proceedings of the 11th International Conference on Learning Representations. Kigali，Rwanda： OpenReview.net, articleTitle=： situated question answering in 3D scenes, refAbstract=null), Reference(id=1322960888369655864, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=1289, pageEnd=1313, url=null, language=null, rfNumber=null, rfOrder=54, authorNames=Maji A, Kumar R, Ghosh A, Anushka, Shah N, Borah A, journalName=null, refType=null, unstructuredReference=MajiA，KumarR，GhoshA，Anushka，ShahN，BorahA，et al. 2025. DRISHTIKON： a multimodal multilingual benchmark for testing language models’ understanding on Indian culture//Proceedings of 2025 Conference on Empirical Methods in Natural Language Processing. Suzhou，China： Association for Computational Linguistics： 1289-1313 ［DOI： 10.18653/v1/2025.emnlp-main.68］, articleTitle=DRISHTIKON： a multimodal multilingual benchmark for testing language models’ understanding on Indian culture, refAbstract=null), Reference(id=1322960888445153337, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=16488, pageEnd=16498, url=null, language=null, rfNumber=null, rfOrder=55, authorNames=Majumdar A, Ajay A, Zhang X H, Putta P, Yenamandra S, Henaff M, journalName=null, refType=null, unstructuredReference=MajumdarA，AjayA，ZhangX H，PuttaP，YenamandraS，HenaffM，et al. 2024. OpenEQA： embodied question answering in the era of foundation models//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Seattle，USA： IEEE： 16488-16498 ［DOI： 10.1109/CVPR52733.2024.01560］, articleTitle=OpenEQA： embodied question answering in the era of foundation models, refAbstract=null), Reference(id=1322960888516456506, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2021, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=56, authorNames=Makoviychuk V, Wawrzyniak L, Guo Y R, Lu M, Storey K, Macklin, M, journalName=Curran Associates Inc, refType=null, unstructuredReference=MakoviychukV，WawrzyniakL，GuoY R，LuM，StoreyK，Macklin，M，et al. 2021. Isaac Gym： high performance GPU-based physics simulation for robot learning//Proceedings of the 35th International Conference on Neural Information Processing Systems. ［s.l.］： Curran Associates Inc., articleTitle=Isaac Gym： high performance GPU-based physics simulation for robot learning, refAbstract=null), Reference(id=1322960890202566715, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2022, volume=7, issue=3, pageStart=7327, pageEnd=7334, url=null, language=null, rfNumber=null, rfOrder=57, authorNames=Mees O, Hermann L, Rosete-Beas E, Burgard W, journalName=IEEE Robotics and Automation Letters, refType=null, unstructuredReference=MeesO，HermannL，Rosete-BeasE and BurgardW. 2022. CALVIN： a benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks. IEEE Robotics and Automation Letters，7（3）： 7327-7334 ［DOI： 10.1109/LRA.2022.3180108］, articleTitle=CALVIN： a benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks, refAbstract=null), Reference(id=1322960890286452796, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=17521, pageEnd=17541, url=null, language=null, rfNumber=null, rfOrder=58, authorNames=Men T Y, Jin Z R, Cao P F, Chen Y B, Liu K, Zhao J, journalName=Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics （Volume 1： Long Papers）, refType=null, unstructuredReference=MenT Y，JinZ R，CaoP F，ChenY B，LiuK and ZhaoJ. 2025. Agent-RewardBench： towards a unified benchmark for reward modeling across perception，planning，and safety in real-world multimodal agents//Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics （Volume 1： Long Papers）. Vienna，Austria： Association for Computational Linguistics： 17521-17541 ［DOI： 10.18653/v1/2025.acl-long.857］, articleTitle=Agent-RewardBench： towards a unified benchmark for reward modeling across perception，planning，and safety in real-world multimodal agents, refAbstract=null), Reference(id=1322960890370338877, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=1, issue=12, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=59, authorNames=Meng Q F, Zhang Y R, Zhang H W, Hu X Y, Luo Y H, journalName=Journal of Image and Graphics, refType=null, unstructuredReference=MengQ F，ZhangY R，ZhangH W，HuX Y and LuoY H. 2025. Research on virtual embodied interaction technology for VR physics experiments. Journal of Image and Graphics，1-12, articleTitle=Research on virtual embodied interaction technology for VR physics experiments, refAbstract=null), Reference(id=1322960890437447742, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=1, pageEnd=12, url=null, language=null, rfNumber=null, rfOrder=60, authorNames=孟启帆, 张怡冉, 张鸿文, 胡晓雁, 骆岩红, journalName=中国图象图形学报, refType=null, unstructuredReference=孟启帆，张怡冉，张鸿文，胡晓雁，骆岩红. 2025. 面向VR物理实验的虚拟具身交互技术研究. 中国图象图形学报，1-12 ［DOI： 10.11834/jig.250425］, articleTitle=面向VR物理实验的虚拟具身交互技术研究, refAbstract=null), Reference(id=1322960890500362303, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.21922.pdf, language=null, rfNumber=null, rfOrder=61, authorNames=Mirjalili V, Giahi R, Kollipara S, Kekuda A, Yao K H, Zhao K, journalName=null, refType=null, unstructuredReference=MirjaliliV，GiahiR，KolliparaS，KekudaA，YaoK H，ZhaoK，et al. 2025. Spatial reasoning in foundation models： benchmarking object-centric spatial understanding ［EB/OL］. ［2025-11-05］., articleTitle=Spatial reasoning in foundation models： benchmarking object-centric spatial understanding, refAbstract=null), Reference(id=1322960890563276864, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2406.02523.pdf, language=null, rfNumber=null, rfOrder=62, authorNames=Nasiriany S, Maddukuri A, Zhang L C, Parikh A, Lo A, Joshi A, journalName=null, refType=null, unstructuredReference=NasirianyS，MaddukuriA，ZhangL C，ParikhA，LoA，JoshiA，et al. 2024. RoboCasa： large-scale simulation of everyday tasks for generalist robots ［EB/OL］. ［2025-11-05］., articleTitle=RoboCasa： large-scale simulation of everyday tasks for generalist robots, refAbstract=null), Reference(id=1322960890638774337, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.15273.pdf, language=null, rfNumber=null, rfOrder=63, authorNames=Ni F, Zhang M, Li P Y, Yuan Y F, Zhang L F, Liu Y C, journalName=null, refType=null, unstructuredReference=NiF，ZhangM，LiP Y，YuanY F，ZhangL F，LiuY C，et al. 2025. Embodied arena： a comprehensive，unified，and evolving evaluation platform for embodied AI ［EB/OL］. ［2025-11-05］., articleTitle=Embodied arena： a comprehensive，unified，and evolving evaluation platform for embodied AI, refAbstract=null), Reference(id=1322960890714271810, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2404.12916.pdf, language=null, rfNumber=null, rfOrder=64, authorNames=Ni Z Y, Ye R, Wei Y X, Xiang Z, Wang Y F, Chen S H, journalName=null, refType=null, unstructuredReference=NiZ Y，YeR，WeiY X，XiangZ，WangY F and ChenS H. 2024. Physical backdoor attack can jeopardize driving with vision-large-language models ［EB/OL］. ［2025-11-05］., articleTitle=Physical backdoor attack can jeopardize driving with vision-large-language models, refAbstract=null), Reference(id=1322960890793963587, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2310.08864.pdf, language=null, rfNumber=null, rfOrder=65, authorNames=O’Neill A, Rehman A, Gupta A, Maddukuri A, Gupta A, Padalkar A, journalName=null, refType=null, unstructuredReference=O’NeillA，RehmanA，GuptaA，MaddukuriA，GuptaA，PadalkarA，et al. 2023. Open X-embodiment： robotic learning datasets and RT-X models ［EB/OL］. ［2025-11-05］., articleTitle=Open X-embodiment： robotic learning datasets and RT-X models, refAbstract=null), Reference(id=1322960890856878148, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=6892, pageEnd=6903, url=null, language=null, rfNumber=null, rfOrder=66, authorNames=O’Neill A, Rehman A, Maddukuri A, Gupta A, Padalkar A, Lee A, journalName=Open X-embodiment： robotic learning datasets and RT-X models： open X-embodiment collaboration0//Proceedings of 2024 IEEE International Conference on Robotics and Automation, refType=null, unstructuredReference=O’NeillA，RehmanA，MaddukuriA，GuptaA，PadalkarA，LeeA，et al. 2024. Open X-embodiment： robotic learning datasets and RT-X models： open X-embodiment collaboration0//Proceedings of 2024 IEEE International Conference on Robotics and Automation. Yokohama，Japan： IEEE： 6892-6903 ［DOI： 10.1109/ICRA57147.2024.10611477］, articleTitle=null, refAbstract=null), Reference(id=1322960890928181317, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2006, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=67, authorNames=Pfeifer R, Bongard J, journalName=null, refType=null, unstructuredReference=PfeiferR and BongardJ. 2006. How the Body Shapes the Way We Think： A New View of Intelligence. Cambridge，USA： The MIT Press, articleTitle=How the Body Shapes the Way We Think： A New View of Intelligence. Cambridge，USA： The MIT Press, refAbstract=null), Reference(id=1322960891007873094, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2506.17221.pdf, language=null, rfNumber=null, rfOrder=68, authorNames=Qi Z Y, Zhang Z X, Yu Y Z, Wang J Q, Zhao H S, journalName=null, refType=null, unstructuredReference=QiZ Y，ZhangZ X，YuY Z，WangJ Q and ZhaoH S. 2025. VLN-R1： vision-language navigation via reinforcement fine-tuning ［EB/OL］. ［2025-11-05］., articleTitle=VLN-R1： vision-language navigation via reinforcement fine-tuning, refAbstract=null), Reference(id=1322960891087564871, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=11948, pageEnd=11956, url=null, language=null, rfNumber=null, rfOrder=69, authorNames=Robey A, Ravichandran Z, Kumar V, Hassani H, Pappas G J, journalName=null, refType=null, unstructuredReference=RobeyA，RavichandranZ，KumarV，HassaniH and PappasG J. 2025. Jailbreaking LLM-controlled robots//Proceedings of 2025 IEEE International Conference on Robotics and Automation （ICRA）. Atlanta，USA： IEEE： 11948-11956 ［DOI： 10.1109/ICRA55743.2025.11128119］, articleTitle=Jailbreaking LLM-controlled robots, refAbstract=null), Reference(id=1322960891171450952, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=645, pageEnd=652, url=null, language=null, rfNumber=null, rfOrder=70, authorNames=Sermanet P, Ding T L, Zhao J, Xia F, Dwibedi D, Gopalakrishnan K, journalName=null, refType=null, unstructuredReference=SermanetP，DingT L，ZhaoJ，XiaF，DwibediD，GopalakrishnanK，et al. 2024. RoboVQA： multimodal long-horizon reasoning for robotics//Proceedings of 2024 IEEE International Conference on Robotics and Automation （ICRA）. Yokohama，Japan： IEEE： 645-652 ［DOI： 10.1109/ICRA57147.2024.10610216］, articleTitle=RoboVQA： multimodal long-horizon reasoning for robotics, refAbstract=null), Reference(id=1322960891246948425, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=3, issue=null, pageStart=null, pageEnd=9150042, url=null, language=null, rfNumber=null, rfOrder=71, authorNames=Sun F C, Chen R F, Ji T Y, Luo Y, Zhou H D, Liu H P, journalName=CAAI Artificial Intelligence Research, refType=null, unstructuredReference=SunF C，ChenR F，JiT Y，LuoY，ZhouH D and LiuH P. 2024. A comprehensive survey on embodied intelligence： advancements，challenges，and future perspectives. CAAI Artificial Intelligence Research，3： #9150042 ［DOI： 10.26599/AIR.2024.9150042］, articleTitle=A comprehensive survey on embodied intelligence： advancements，challenges，and future perspectives, refAbstract=null), Reference(id=1322960891322445898, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2410.00425.pdf, language=null, rfNumber=null, rfOrder=72, authorNames=Tao S, Xiang F B, Shukla A, Qin Y Z, Hinrichsen X, Yuan X D, journalName=null, refType=null, unstructuredReference=TaoS，XiangF B，ShuklaA，QinY Z，HinrichsenX，YuanX D，et al. 2024. ManiSkill3： GPU parallelized robotics simulation and rendering for generalizable embodied AI ［EB/OL］. ［2025-11-05］., articleTitle=ManiSkill3： GPU parallelized robotics simulation and rendering for generalizable embodied AI, refAbstract=null), Reference(id=1322960891389554763, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2011, volume=null, issue=null, pageStart=1521, pageEnd=1528, url=null, language=null, rfNumber=null, rfOrder=73, authorNames=Torralba A, Efros A A, journalName=Proceedings of the CVPR 2011, refType=null, unstructuredReference=TorralbaA and EfrosA A. 2011. Unbiased look at dataset bias//Proceedings of the CVPR 2011. Colorado Springs，USA： IEEE： 1521-1528 ［DOI： 10.1109/CVPR.2011.5995347］, articleTitle=Unbiased look at dataset bias, refAbstract=null), Reference(id=1322960891477635148, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2015, volume=null, issue=null, pageStart=1137, pageEnd=1143, url=null, language=null, rfNumber=null, rfOrder=74, authorNames=Ushani A K, Carlevaris-Bianco N, Cunningham A G, Galceran E, Eustice R M, journalName=null, refType=null, unstructuredReference=UshaniA K，Carlevaris-BiancoN，CunninghamA G，GalceranE and EusticeR M. 2015. Continuous-time estimation for dynamic obstacle tracking//Proceedings of 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Hamburg，Germany： IEEE： 1137-1143 ［DOI： 10.1109/IROS.2015.7353513］, articleTitle=Continuous-time estimation for dynamic obstacle tracking, refAbstract=null), Reference(id=1322960891544744013, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024a, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2407.10943.pdf, language=null, rfNumber=null, rfOrder=75, authorNames=Wang H Q, Chen J H, Huang W S, Ben Q W, Wang T, Mi B Y, journalName=null, refType=null, unstructuredReference=WangH Q，ChenJ H，HuangW S，BenQ W，WangT，MiB Y，et al. 2024a. GRUtopia： dream general robots in a city at scale ［EB/OL］. ［2025-11-05］., articleTitle=GRUtopia： dream general robots in a city at scale, refAbstract=null), Reference(id=1322960891624435790, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024b, volume=null, issue=null, pageStart=#3952, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=76, authorNames=Wang L R, Chen X L, Zhao J L, He K M, journalName=null, refType=null, unstructuredReference=WangL R，ChenX L，ZhaoJ L and HeK M. 2024b. Scaling proprioceptive-visual learning with heterogeneous pre-trained transformers//Proceedings of the 38th International Conference on Neural Information Processing Systems. Vancouver，Canada： Curran Associates Inc.： #3952, articleTitle=Scaling proprioceptive-visual learning with heterogeneous pre-trained transformers, refAbstract=null), Reference(id=1322960891687350351, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=1615, pageEnd=1638, url=null, language=null, rfNumber=null, rfOrder=77, authorNames=Wang Z J, Zhou Z H, Song J Y, Huang Y H, Shu Z, Ma L, journalName=（FSE）, refType=null, unstructuredReference=WangZ J，ZhouZ H，SongJ Y，HuangY H，ShuZ and MaL. 2025. VLATest： testing and evaluating vision-language-action models for robotic manipulation. Proceedings of the ACM on Software Engineering，2（FSE）： 1615-1638 ［DOI： 10.1145/3729343］, articleTitle=VLATest： testing and evaluating vision-language-action models for robotic manipulation. Proceedings of the ACM on Software Engineering，2, refAbstract=null), Reference(id=1322960891762847824, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2412.13877.pdf, language=null, rfNumber=null, rfOrder=78, authorNames=Wu K, Hou C K, Liu J M, Che Z P, Ju X Z, Yang Z Q, journalName=null, refType=null, unstructuredReference=WuK，HouC K，LiuJ M，CheZ P，JuX Z，YangZ Q，et al. 2024. RoboMIND： benchmark on multi-embodiment intelligence normative data for robot manipulation ［EB/OL］. ［2025-11-05］., articleTitle=RoboMIND： benchmark on multi-embodiment intelligence normative data for robot manipulation, refAbstract=null), Reference(id=1322960891838345297, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.14404.pdf, language=null, rfNumber=null, rfOrder=79, authorNames=Wu X C, Liu J X, Huang D L, Wang Y F, Shi Y Y, Chen K D, journalName=null, refType=null, unstructuredReference=WuX C，LiuJ X，HuangD L，WangY F，ShiY Y，ChenK D，et al. 2025. ViC-bench： benchmarking visual-interleaved chain-of-thought capability in MLLMs with free-style intermediate state representations ［EB/OL］. ［2025-11-05］., articleTitle=ViC-bench： benchmarking visual-interleaved chain-of-thought capability in MLLMs with free-style intermediate state representations, refAbstract=null), Reference(id=1322960891909648466, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025a, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2508.06553.pdf, language=null, rfNumber=null, rfOrder=80, authorNames=Xiao J H, Yan B W, Zhang J B, Wang J, Li C Y, Cheng Z X, journalName=null, refType=null, unstructuredReference=XiaoJ H，YanB W，ZhangJ B，WangJ，LiC Y，ChengZ X，et al. 2025a. Data assessment for embodied intelligence ［EB/OL］. ［2025-11-05］., articleTitle=Data assessment for embodied intelligence, refAbstract=null), Reference(id=1322960891985145939, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025b, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2508.06553.pdf, language=null, rfNumber=null, rfOrder=81, authorNames=Xiao J H, Zhang J B, Yan B W, Guo S Y, Ye T R, Zhang K W, journalName=null, refType=null, unstructuredReference=XiaoJ H，ZhangJ B，YanB W，GuoS Y，YeT R，ZhangK W，et al. 2025b. Static and plugged： make embodied evaluation simple［EB/OL］. ［2025-11-05］., articleTitle=Static and plugged： make embodied evaluation simple, refAbstract=null), Reference(id=1322960892052254804, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2502.11426.pdf, language=null, rfNumber=null, rfOrder=82, authorNames=Xu T, Pan C, Rao M B, Datar A, Pokhrel A, Lu Y, journalName=null, refType=null, unstructuredReference=XuT，PanC，RaoM B，DatarA，PokhrelA，LuY，et al，2025. Verti-bench： a general and scalable off-road mobility benchmark for vertically challenging terrain ［EB/OL］. ［2025-11-05］., articleTitle=Verti-bench： a general and scalable off-road mobility benchmark for vertically challenging terrain, refAbstract=null), Reference(id=1322960892123557973, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2509.25794.pdf, language=null, rfNumber=null, rfOrder=83, authorNames=Xue H, Ge Y, Zeng Y, Li Z, Liu M Y, Chen Y, journalName=null, refType=null, unstructuredReference=XueH，GeY，ZengY，LiZ，LiuM Y，ChenY，et al. 2025. Point-it-out： benchmarking embodied reasoning for vision language models in multi-stage visual grounding ［EB/OL］. ［2025-11-05］., articleTitle=Point-it-out： benchmarking embodied reasoning for vision language models in multi-stage visual grounding, refAbstract=null), Reference(id=1322960892186472534, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2412.07215.pdf, language=null, rfNumber=null, rfOrder=84, authorNames=Yan F, Liu F F, Zheng L M, Zhong Y F, Huang Y Y, Guan Z C, journalName=null, refType=null, unstructuredReference=YanF，LiuF F，ZhengL M，ZhongY F，HuangY Y，GuanZ C，et al. 2024. RoboTron-Mani： all-in-one multimodal large model for robotic manipulation ［EB/OL］. ［2025-11-05］., articleTitle=RoboTron-Mani： all-in-one multimodal large model for robotic manipulation, refAbstract=null), Reference(id=1322960892266164311, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2412.14171.pdf, language=null, rfNumber=null, rfOrder=85, authorNames=Yang J, Yang S, Gupta A W, Han R, Fei-Fei L, Xie S, journalName=null, refType=null, unstructuredReference=YangJ，YangS，GuptaA W，HanR，Fei-FeiL and XieS. 2025. Thinking in space： how multimodal large language models see，remember，and recall spaces ［EB/OL］. ［2025-11-05］., articleTitle=Thinking in space： how multimodal large language models see，remember，and recall spaces, refAbstract=null), Reference(id=1322960892358439000, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2412.13178.pdf, language=null, rfNumber=null, rfOrder=86, authorNames=Yin S, Pang X H, Ding Y Z, Chen M L, Bi Y T, Xiong Y C, journalName=null, refType=null, unstructuredReference=YinS，PangX H，DingY Z，ChenM L，BiY T，XiongY C，et al. 2024. SafeAgentBench： a benchmark for safe task planning of embodied LLM agents ［EB/OL］. ［2025-11-05］., articleTitle=SafeAgentBench： a benchmark for safe task planning of embodied LLM agents, refAbstract=null), Reference(id=1322960892433936473, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2506.14697.pdf, language=null, rfNumber=null, rfOrder=87, authorNames=Ying Z H, Wang L, Xiao Y S, Wang J K, Ma Y Q, Guo J Y, journalName=null, refType=null, unstructuredReference=YingZ H，WangL，XiaoY S，WangJ K，MaY Q，GuoJ Y，et al. 2025. AGENTSAFE： benchmarking the safety of embodied agents on hazardous instructions ［EB/OL］. ［2025-11-05］., articleTitle=AGENTSAFE： benchmarking the safety of embodied agents on hazardous instructions, refAbstract=null), Reference(id=1322960892501045338, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2019, volume=null, issue=null, pageStart=1094, pageEnd=1100, url=null, language=null, rfNumber=null, rfOrder=88, authorNames=Yu T H, Quillen D, He Z P, Julian R, Hausman K, Finn C, journalName=null, refType=null, unstructuredReference=YuT H，QuillenD，HeZ P，JulianR，HausmanK，FinnC，et al. 2019. Meta-world： a benchmark and evaluation for multi-task and meta reinforcement learning//Proceedings of the 3rd Annual Conference on Robot Learning. Osaka，Japan： PMLR： 1094-1100, articleTitle=Meta-world： a benchmark and evaluation for multi-task and meta reinforcement learning, refAbstract=null), Reference(id=1322960892563959899, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.08548.pdf, language=null, rfNumber=null, rfOrder=89, authorNames=Yuan Y F, Cui H Q, Chen Y B, Dong Z B, Ni F, Kou L X, journalName=null, refType=null, unstructuredReference=YuanY F，CuiH Q，ChenY B，DongZ B，NiF，KouL X，et al. 2025. From seeing to doing： bridging reasoning and decision for robotic manipulation ［EB/OL］. ［2025-11-05］., articleTitle=From seeing to doing： bridging reasoning and decision for robotic manipulation, refAbstract=null), Reference(id=1322960892626874460, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024d, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2407.20242.pdf, language=null, rfNumber=null, rfOrder=90, authorNames=Zhang H, Zhu C, Wang X, Zhou Z, Hu S and Zhang L Y, journalName=null, refType=null, unstructuredReference=ZhangH，ZhuC，WangX，ZhouZ，Hu S and ZhangL Y，2024d. BadRobot： jailbreaking LLM-based embodied AI in the physical world ［EB/OL］. ［2025-11-05］., articleTitle=BadRobot： jailbreaking LLM-based embodied AI in the physical world, refAbstract=null), Reference(id=1322960892702371933, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024a, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2412.18774.pdf, language=null, rfNumber=null, rfOrder=91, authorNames=Zhang J B, Li C Y, Hao J, Jia J, Duan H Y, Zheng G Q, journalName=null, refType=null, unstructuredReference=ZhangJ B，LiC Y，HaoJ，JiaJ，DuanH Y，ZhengG Q，et al. 2024a. Embodied image quality assessment for robotic intelligence［EB/OL］. ［2025-11-05］., articleTitle=Embodied image quality assessment for robotic intelligence, refAbstract=null), Reference(id=1322960892769480798, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024b, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2412.06224.pdf, language=null, rfNumber=null, rfOrder=92, authorNames=Zhang J Z, Wang K Y, Wang S A, Li M H, Liu H R, Wei S L, journalName=null, refType=null, unstructuredReference=ZhangJ Z，WangK Y，WangS A，LiM H，LiuH R，WeiS L，et al. 2024b. Uni-NaVid： a video-based vision-language-action model for unifying embodied navigation tasks ［EB/OL］. ［2025-11-05］., articleTitle=Uni-NaVid： a video-based vision-language-action model for unifying embodied navigation tasks, refAbstract=null), Reference(id=1322960892844978271, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024c, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=null, rfOrder=93, authorNames=Zhang J Z, Wang K Y, Xu R T, Zhou G Z, Hong Y C, Fang X M, journalName=Robotics： Science and Systems XX, refType=null, unstructuredReference=ZhangJ Z，WangK Y，XuR T，ZhouG Z，HongY C，FangX M，et al. 2024c. NaVid： video-based VLM plans the next step for vision-and-language navigation//Robotics： Science and Systems XX. Delft，the Netherlands： ［s.n.］, articleTitle=NaVid： video-based VLM plans the next step for vision-and-language navigation, refAbstract=null), Reference(id=1322960892933058656, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025a, volume=null, issue=null, pageStart=21566, pageEnd=21573, url=null, language=null, rfNumber=null, rfOrder=94, authorNames=Zhang L F, Wang Y N, Gu H J, Hamidizadeh A, Zhang Z G, Liu Y C, journalName=null, refType=null, unstructuredReference=ZhangL F，WangY N，GuH J，HamidizadehA，ZhangZ G，LiuY C，et al. 2025a. ET-plan-bench： embodied task-level planning benchmark towards spatial-temporal cognition with foundation models//Proceedings of 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Hangzhou，China： IEEE： 21566-21573 ［DOI： 10.1109/IROS60139.2025.11246658］, articleTitle=ET-plan-bench： embodied task-level planning benchmark towards spatial-temporal cognition with foundation models, refAbstract=null), Reference(id=1322960893000167521, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025b, volume=null, issue=null, pageStart=11142, pageEnd=11152, url=null, language=null, rfNumber=null, rfOrder=95, authorNames=Zhang S D, Xu Z, Liu P J, Yu X P, Li Y, Gao Q H, journalName=null, refType=null, unstructuredReference=ZhangS D，XuZ，LiuP J，YuX P，LiY，GaoQ H，et al. 2025b. VLABench： a large-scale benchmark for language-conditioned robotics manipulation with long-horizon reasoning tasks//Proceedings of 2025 IEEE/CVF International Conference on Computer Vision. ［s.l.］： ［s.n.］： 11142-11152, articleTitle=VLABench： a large-scale benchmark for language-conditioned robotics manipulation with long-horizon reasoning tasks, refAbstract=null), Reference(id=1322960894640140386, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025c, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2503.11089.pdf, language=null, rfNumber=null, rfOrder=96, authorNames=Zhang Y, Zhang Q, Ju X Z, Liu Z Y, Mao J L, Sun J K, journalName=null, refType=null, unstructuredReference=ZhangY，ZhangQ，JuX Z，LiuZ Y，MaoJ L，SunJ K，et al. 2025c. EmbodiedVSR： dynamic scene graph-guided chain-of-thought reasoning for visual spatial tasks ［EB/OL］. ［2025-11-05］., articleTitle=EmbodiedVSR： dynamic scene graph-guided chain-of-thought reasoning for visual spatial tasks, refAbstract=null), Reference(id=1322960894719832163, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2026, volume=91, issue=null, pageStart=null, pageEnd=103255, url=null, language=null, rfNumber=null, rfOrder=97, authorNames=Zhang Z C, Wang J Y, Guo Y J, Wen F R, Chen Z J, Wang H Q, journalName=Displays, refType=null, unstructuredReference=ZhangZ C，WangJ Y，GuoY J，WenF R，ChenZ J，WangH Q，et al. 2026. AIBench： towards trustworthy evaluation under the 45°law. Displays，91： #103255 ［DOI： 10.1016/j.displa.2025.103255］, articleTitle=AIBench： towards trustworthy evaluation under the 45°law, refAbstract=null), Reference(id=1322960894795329636, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025d, volume=68, issue=12, pageStart=null, pageEnd=221301, url=null, language=null, rfNumber=null, rfOrder=98, authorNames=Zheng Z C, Wang J Y, Wen F R, Guo Y J, Zhao X Y, Fang X Y, journalName=Science China Information Sciences, refType=null, unstructuredReference=ZhengZ C，WangJ Y，WenF R，GuoY J，ZhaoX Y，FangX Y，et al. 2025d. Large multimodal models evaluation： a survey. Science China Information Sciences，68（12）： #221301 ［DOI： 10.1007/s11432-025-4676-4］, articleTitle=Large multimodal models evaluation： a survey, refAbstract=null), Reference(id=1322960894866632805, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2407.06951.pdf, language=null, rfNumber=null, rfOrder=99, authorNames=Zheng L M, Yan F, Liu F F, Feng C J, Kang Z L, Ma L, journalName=null, refType=null, unstructuredReference=ZhengL M，YanF，LiuF F，FengC J，KangZ L and MaL. 2024. RoboCAS： a benchmark for robotic manipulation in complex object arrangement scenarios ［EB/OL］. ［2025-11-05］., articleTitle=RoboCAS： a benchmark for robotic manipulation in complex object arrangement scenarios, refAbstract=null), Reference(id=1322960894937935974, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2505.12084.pdf, language=null, rfNumber=null, rfOrder=100, authorNames=Zhong N H, Caro S, Iskandar A, Ramesh M, Smith S L, journalName=null, refType=null, unstructuredReference=ZhongN H，CaroS，IskandarA，RameshM and SmithS L. 2025. Bench-NPIN： benchmarking non-prehensile interactive navigation［EB/OL］. ［2025-11-05］., articleTitle=Bench-NPIN： benchmarking non-prehensile interactive navigation, refAbstract=null), Reference(id=1322960895005044839, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025a, volume=null, issue=null, pageStart=6940, pageEnd=6949, url=null, language=null, rfNumber=null, rfOrder=101, authorNames=Zhou W J, Tao M L, Zhao C Y, Guo H Y, Dong H H, Tang M, journalName=null, refType=null, unstructuredReference=ZhouW J，TaoM L，ZhaoC Y，GuoH Y，DongH H，TangM，et al. 2025a. PhysVLM： enabling visual language models to understand robotic physical reachability//Proceedings of 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Nashville，USA： 6940-6949 ［DOI： 10.1109/CVPR52734.2025.00651］, articleTitle=PhysVLM： enabling visual language models to understand robotic physical reachability, refAbstract=null), Reference(id=1322960895093125224, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2025b, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2503.24278.pdf, language=null, rfNumber=null, rfOrder=102, authorNames=Zhou Z Y, Atreya P, Tan Y L, Pertsch K, Levine S, journalName=null, refType=null, unstructuredReference=ZhouZ Y，AtreyaP，TanY L，PertschK and LevineS. 2025b. AutoEval： autonomous evaluation of generalist robot manipulation policies in the real world ［EB/OL］. ［2025-11-05］., articleTitle=AutoEval： autonomous evaluation of generalist robot manipulation policies in the real world, refAbstract=null), Reference(id=1322960895151845481, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2024, volume=null, issue=null, pageStart=null, pageEnd=null, url=https://arxiv.org/pdf/2408.04449.pdf, language=null, rfNumber=null, rfOrder=103, authorNames=Zhu Z H, Wu B Z, Zhang Z Y, Han L, Liu Q S, Wu B Y, journalName=null, refType=null, unstructuredReference=ZhuZ H，WuB Z，ZhangZ Y，HanL，LiuQ S and WuB Y. 2024. EARBench： towards evaluating physical risk awareness for task planning of foundation model-based embodied AI agents ［EB/OL］. ［2025-11-05］., articleTitle=EARBench： towards evaluating physical risk awareness for task planning of foundation model-based embodied AI agents, refAbstract=null), Reference(id=1322960895218954346, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, doi=null, pmid=null, pmcid=null, year=2023, volume=null, issue=null, pageStart=2165, pageEnd=2183, url=null, language=null, rfNumber=null, rfOrder=104, authorNames=Zitkovich B, Yu T H, Xu S C, Xu P, Xiao T, Xia F, journalName=null, refType=null, unstructuredReference=ZitkovichB，YuT H，XuS C，XuP，XiaoT，XiaF，et al. 2023. RT-2： vision-language-action models transfer web knowledge to robotic control//Proceedings of 2023 Conference on Robot Learning. Atlanta，USA： PMLR： 2165-2183, articleTitle=RT-2： vision-language-action models transfer web knowledge to robotic control, refAbstract=null)], funds=[Fund(id=1322960882950615040, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, awardId=null, language=EN, fundingSource=National Natural Science Foundation of China（62225112，625B2118）；Singapore MoE Foundation（MOE-T2EP20123-0006）, fundOrder=null, country=null), Fund(id=1322960883005140993, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, awardId=null, language=CN, fundingSource=国家自然科学基金项目（62225112，625B2118）；新加坡教育部基金项目（MOE-T2EP20123-0006）, fundOrder=null, country=null)], companyList=[AuthorCompany(id=1322960870430618526, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=1, ext=[AuthorCompanyExt(id=1322960870459978655, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870430618526, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Center of AI Evaluation，Shanghai AI Lab，Shanghai 200032，China), AuthorCompanyExt(id=1322960870476755872, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870430618526, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 上海人工智能实验室评测专项组，上海;200032)]), AuthorCompany(id=1322960870548059041, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=2, ext=[AuthorCompanyExt(id=1322960870560641954, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870548059041, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Information Science and Electronic Engineering，Shanghai Jiao Tong University，Shanghai 200240，China), AuthorCompanyExt(id=1322960870573224867, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870548059041, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 上海交通大学信息与电子工程学院，上海 200240)]), AuthorCompany(id=1322960870644528036, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=3, ext=[AuthorCompanyExt(id=1322960870652916645, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870644528036, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 Institute of Data and Information，Tsinghua Shenzhen International Graduate School，Shenzhen 518055，China), AuthorCompanyExt(id=1322960870661305254, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960870644528036, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 清华大学深圳国际研究生院数据与信息研究院，深圳 518055)]), AuthorCompany(id=1322960872326443943, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=4, ext=[AuthorCompanyExt(id=1322960872339026856, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960872326443943, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 Yuanpei College，Peking University，Beijing 100871，China), AuthorCompanyExt(id=1322960872355804073, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960872326443943, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=4 北京大学元培学院，北京 100871)]), AuthorCompany(id=1322960872620045226, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=5, ext=[AuthorCompanyExt(id=1322960872632628139, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960872620045226, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=5 Department of Psychology，Tongji University，Shanghai 200292，China), AuthorCompanyExt(id=1322960872649405356, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960872620045226, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=5 同济大学应用心理学系，上海 200292)]), AuthorCompany(id=1322960872750068653, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, xref=6, ext=[AuthorCompanyExt(id=1322960872758457262, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960872750068653, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=6 College of Computing and Data Science，Nanyang Technological University，Singapore 639798，Singapore), AuthorCompanyExt(id=1322960872766845871, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, companyId=1322960872750068653, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=6 南洋理工大学计算机与数据科学学院，新加坡 639798，新加坡)])], figs=[ArticleFig(id=1322960879800693739, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, label=Fig.1, caption=Perception-cognition-decision-execution loop for intelligence systems, figureFileSmall=1ryRKUlPwOfuJtH5zVAqIw==, figureFileBig=R/Vypy3oOwNxRDL3pDE9DA==, tableContent=null), ArticleFig(id=1322960881407112174, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, label=图1, caption=智能系统的感知—认知—决策—执行闭环, figureFileSmall=1ryRKUlPwOfuJtH5zVAqIw==, figureFileBig=R/Vypy3oOwNxRDL3pDE9DA==, tableContent=null), ArticleFig(id=1322960881600050159, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, label=Fig.2, caption=Scatter plot of Octo VLA model performance on 50 task samples, figureFileSmall=Y9LBvB+KdYMwjwB5DgiESQ==, figureFileBig=Ymz/hYXt9QrZTN6xtwK4pw==, tableContent=null), ArticleFig(id=1322960881675547632, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=CN, label=图2, caption=Octo 模型在50个任务样本上的性能散点图, figureFileSmall=Y9LBvB+KdYMwjwB5DgiESQ==, figureFileBig=Ymz/hYXt9QrZTN6xtwK4pw==, tableContent=null), ArticleFig(id=1322960881742656497, tenantId=1146029695717560320, journalId=1249024232475115590, articleId=1322960864705393559, language=EN, label=Table 1, caption=Ability evaluation dimensions for human and embodied intelligence, figureFileSmall=null, figureFileBig=null, tableContent=
5 Department of Psychology，Tongji University，Shanghai 200292，China
6 College of Computing and Data Science，Nanyang Technological University，Singapore 639798，Singapore
出版时间: 2026-08-16
doi: 10.11834/jig.250550
文章导航
摘要
收起
具身智能通过“身体—环境—任务”闭环交互，被视为迈向通用人工智能的核心路径。然而，与模型参数和训练数据的高速扩张形成鲜明对照，其评测体系仍处于“任务碎片化、指标多元化、平台封闭化”的自发状态，造成同一功能在不同研究中的性能差异超过10%，却无法判定来源是算法创新、数据扩充还是评测偏差，严重阻碍了技术迭代与产业落地。本文围绕“从感知到执行”的完整链路，对2020—2025年间发表于CVPR（IEEE/CVF Conference on Computer Vision and Pattern Recognition）、ICRA（IEEE International Conference on Robotics and Automation）、NeurIPS（Conference on Neural Information Processing Systems）等顶级会议与 Nature、IJRR（International Journal of Robotic Research）、JMLR（Journal of Machine Learning Research）等期刊的百余篇文献进行系统梳理，首次提出“静态数据集—仿真平台—真实机器人”三级金字塔式评测范式，并从感知、认知、决策、执行4个环节拆解出20余项核心能力维度与量化指标，分别从方法学假设、适用边界、固有局限和成本—可信度曲线4个角度进行横向对比。本文进一步汇总了46套主流基在数据规模、任务类型、评估指标、开源程度和安全伦理考量等维度的差异。基于此，提出“能力导向、协议统一、三级协同”的未来框架：1）在任务层，由“功能对标”转向“能力对标”，建立可分解、可溯源、可加权的多维误差体系；2）在协议层，制定统一的场景描述、接口规范与指标定义，实现跨平台、跨任务、跨模型的可比性；3）在系统层，构建可远程接入、7 × 24 h运行的共享机器人集群，形成“仿真预训练—真机微调—在线更新”的闭环追踪，降低重复建设成本，打通从实验室到场景落地的“最后一公里”。最后讨论了安全伦理、文化偏见和能效评估等新维度如何纳入量化框架，并给出标准化路线图的短、中、长期目标。本文工作为具身智能从“技术涌现”走向“科学共识”提供了可操作的评测基础设施与参考范式，具体的静态—仿真—真机榜单展示在：https://opencompass.org.cn/embodied-intelligence。
Embodied intelligence refers to agents that sense，reason，act，and learn through a tight “body-environment-task” feedback loop. It is widely regarded as the most promising route to general-purpose artificial intelligence （AI）. Paradoxically，although model parameters，training data，and compute budgets have grown by four orders of magnitude in only five years，the evaluation ecosystem remains fragmented and largely irreproducible. The same functional capability （e.g.，“pick-and-place”） is formalized as object-detection mean average precision in one paper，as 3D intersection over union in a second，as task success rate in a third，and as human preference score in a fourth，with absolute gaps that exceed 10%—15% among “state-of-the-art” results. Environments，random seeds，physics parameters，sensor noise models，and success criteria are rarely open-sourced，and thus，the community cannot tell whether an apparent improvement originates from algorithmic innovation，data scale，evaluation cherry picking，or simple benchmark over-fitting. This uncertainty has become a critical bottleneck for scientific progress and industrial adoption. To address the gap，we conduct a systematic survey of 100+ papers published in IEEE/CVF Conference on Computer Vision and Pattern Recognition，IEEE International Conference on Robotics and Automation，Conference on Neural Information Processing Systems，Robotics： Science and Systems，International Conference on Learning Representations，International Journal of Robotic Research，and Journal of Machine Learning Research between 2020 and 2025. Then，we present the first holistic analysis of embodied AI evaluation. We organize the landscape into a three-tier pyramid： static datasets，simulation benchmarks，and real-robot trials. Each tier trades off cost，controllability，and ecological validity. Across all tiers，we decompose evaluation space into four tightly coupled stages： perception，cognition，decision，and execution. Then，we identify fine-grained capability dimensions along with 20+ quantitative metrics. For every dimension，we analyze the following： 1） the underlying methodological assumptions，2） the boundary conditions under which the metric is valid，3） the intrinsic limitations （e.g.，sim-to-real physics mismatch，dataset bias，human-label variance），and 4） the cost-credibility curve that determines when a researcher should ascend from static to simulation，and finally to the real world.All evaluations in this study are based on a theoretical “perception-cognition-decision-execution loop” framework. In embodied AI，perception is the process by which an agent converts raw，asynchronous，and often noisy signals from heterogeneous physical sensors，such as RGB-D cameras，LiDAR，tactile skins，joint encoders，force/torque，and inertial measurement unit units，into a registered，time-aligned digital twin that supplies metric 3D geometry and semantic labels. Cognition is then built on this torrent to distill structured，memory-enabled，and generalizable world models that encode object affordances，physical common sense，and relational and causal links，grounding free-form human language into the resulting latent space. Decision-making combines these world models with an explicit task goal and searches across symbolic，kinematic，and dynamic levels to output a coherent，risk-aware，and resource-efficient plan that is expressed as a hierarchy of subgoals，trajectories，and low-level set points. Finally，execution transforms this plan into real-world mechanical energy through high-rate feedback controllers that track desired motions or forces while compensating for actuator nonlinearities，shock，thermal limits，and human safety constraints，closing the perception-cognition-decision-execution loop that defines embodied AI. In contrast with humans，whose perception-cognition-decision-execution loop is seamlessly integrated，currently available embodied agents suffer structural flaws in every stage. Consequently，evaluation must decouple the four blocks and score accuracy，error，and latency separately to pinpoint bottlenecks. Static tests，i.e.，“written exams”，run on a GPU with a few megabytes of images or point clouds in minutes，but erase dynamics and physics. Simulation，i.e.，the “job interview”，adds interactive physics in tens of seconds but still drifts away from reality. Real robot trials，i.e.，“practical tests”，cost thousands，take days to deploy and minutes per run，but expose every sensing drift，prediction error，or control delay，offering the ultimate ground truth before field deployment. Therefore，by integrating the three-tier pyramid of static test-simulation-real robot evaluation into the closed perception-cognition-decision-execution loop，this work delivers the first systematic quantification of coverage，cost，and accuracy trade-offs across more than 100 embodied AI benchmarks，establishing a reproducible，traceable，and extensible evaluation baseline that not only exposes the exact stage at which state-of-the-art systems fail，but also provides a principled road map for allocating research budgets，standardizing protocols，and certifying safety before real-world deployment，marking a pivotal transition from empirical competition to scientific measurement，and accelerating the field’s development toward trustworthy，deployable，and general-purpose embodied intelligence. The detailed static-dynamic-real embodied intelligence evaluation can be accessed at https://opencompass.org.cn/embodied-intelligence.
具身智能与传统人工智能在智能获取方式上存在本质区别。传统人工智能，尤其是在大数据和深度学习驱动下取得巨大成功的领域，如计算机视觉（computer vision，CV）和自然语言处理（natural language processing，NLP），其模型训练主要依赖于大规模、预先收集并标注的静态数据集（LeCun等，2015）。这种范式使得模型能够在特定、受限的环境中表现出色，但一旦面临开放、动态和非结构化的真实世界，其泛化能力和鲁棒性便会受到严峻挑战（Torralba和Efros，2011）。相比之下，具身智能体通过与环境的实时互动来获取训练数据、检验假设并优化行为策略（Kaelbling等，1996）。这种学习方式使得智能体能够持续适应环境变化，并从交互中自主发现和学习新的概念与技能。
近年来，随着算法、理论和硬件的飞速发展，具身智能已被广泛认为是探索通用人工智能（artificial general intelligence，AGI）的核心路径，获得了学术界和工业界的一致共识（Lake等，2017）。AGI 的目标是创造出能够像人类一样思考、学习和解决各种复杂问题的智能系统。许多研究者认为，要实现这一目标，智能体必须拥有一个物理身体，使其能够像人类一样通过感知和行动与物理世界进行深度交互。这种交互不仅是获取信息的手段，更是智能形成的源泉。通过具身交互，智能体可以学习到关于物理世界的基本规律，积累常识知识，并发展出解决新问题的能力。因此，具身智能被视为弥合当前AI与AGI之间鸿沟的关键，它迫使智能体必须处理真实世界的复杂性和不确定性，从而催生出更鲁棒、更灵活和更具适应性的智能。
具身智能的发展主要围绕两大核心任务——操作（manipulation）和导航（navigation）。早期研究聚焦于特定任务和环境下训练的策略模型，形成了以强化学习（reinforcement learning，RL）和模仿学习（imitation learning，IL）为主的学习方法。随着大语言模型（large language model，LLM）和多模态大模型（multimodal large language model，MLLM）的崛起，出现了将大模型引入具身智能的具身大模型（embodied large model，ELM）范式。具身智能任务通常可形式化为部分可观测马尔可夫决策过程（partially observable Markov decision process，POMDP）。在现实环境中，智能体无法直接访问完整的环境状态，而只能通过传感器获得部分观测，因此 POMDP 成为描述具身交互的自然框架。一个典型的 POMDP 可表示为七元组
在认知层级，关注更高层次的语义和推理能力，一般以视觉问答（VQA）形式出现。Can-Do在多模态复杂场景下测试理解、推理与高阶规划能力，结合Acc和BLEU指标评估任务通用性（Chia等，2024）。RoboVQA构建大规模视频—文本问答数据集，检验模型在长时视觉推理任务中的语言生成精度（Sermanet等，2024）。OpenEQA与UniEQA代表开放词汇问答类基准，前者以Acc和GPT评估语言理解精度（Majumdar等，2024），后者通过Uniscore、Acc和BLEU综合衡量理解与生成一致性（Zhang等，2025b）。VSI-Bench聚焦视频语义整合，使用MRA和Recall@K衡量跨帧语义一致性及时空关系理解能力（Yang等，2025）。ERQA侧重物理交互因果推理，采用Acc、Recall和F1分数评估因果推理能力（Gu等，2025）。PhyBlock结合物理常识问答任务，通过Acc、precision/recall及mAP（mean average precision）衡量隐式规律抽象能力（Ma等，2025）。MineAnyBuild以Minecraft建造场景评估语言指令与生成方案一致性，采用GPT评分和指令执行匹配率（Exec Rate）量化规划与推理能力（Chen等，2025c）。Embodied Arena、StaticEmbodiedBench和UniEQA在认知层提供跨模态与多任务理解指标，从语言维度描绘具身智能系统的“语义理解—因果推理—生成一致性”链条（Ni等，2025；Xiao等，2025b；Zhang等，2024b）。
为了验证静态关键帧评测方法在预测真实机器人执行任务表现中的有效性，在UR5机械臂上进行了测试，并采用Octo 模型进行任务预测与执行。实验选取了50个基础任务实例，涵盖5类基础任务类型，包括拾取、放置、按压、拉和推。实验中采用两种评分机制：真实执行评分和静态评测评分。对于真实执行评分，机器人尝试完成任务，任务在以下任一条件下终止：任务成功完成、达到最大步数限制（50 步）或触发安全停止机制（如碰撞）。每个任务实例进行了50轮测试，以获取最终状态下的误差值，并据此计算位置、旋转和夹爪开合3个维度的动态评测分数。静态评测评分则通过关键帧方法获得，即由 Octo 模型输出与目标状态进行比较，从而直接得到静态分数，无需真实机器人执行操作。
Gemini robotics： bringing AI into the physical world
AbeyruwanS，AinslieJ，AlayracJ B，ArenasM G，ArmstrongT，BalakrishnaA，et al. 2025. Gemini robotics： bringing AI into the physical world ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2503.20020.pdf
Embodied intelligence： recent advances and future perspectives
AnZ L，YuX Q，WangC，ZhangJ L，LiY and SongC H. 2025. Embodied intelligence： recent advances and future perspectives. The Innovation Informatics，1（1）： #100008 ［DOI： 10.59717/j.xinn-inform.2025.100008］
Embodied cognition： a field guide
AndersonM L. 2003. Embodied cognition： a field guide. Artificial Intelligence，149（1）： 91-130 ［DOI： 10.1016/S0004-3702（03）00054-7］
Scan Q A： 3D question answering for spatial scene understanding
AzumaD，MiyanishiT，KuritaS and KawanabeM. 2022. Scan Q A： 3D question answering for spatial scene understanding//Proceedings of 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans，USA： IEEE： 19107-19117
Cosmos-Reason1： from physical common sense to embodied reasoning
AzzoliniA，BaiJ J，BrandonH，CaoJ X，ChattopadhyayP，ChenH Y，et al. 2025. Cosmos-Reason1： from physical common sense to embodied reasoning ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2503.15558.pdf
Can you SPLICE it together？ A human curated benchmark for probing visual reasoning in VLMs
BalloutM，WilfredO，YaghoubiS，AbdelmoneimN M，MayerJ and BruniE. 2025. Can you SPLICE it together？ A human curated benchmark for probing visual reasoning in VLMs//Findings of the Association for Computational Linguistics： EMNLP 2025. Suzhou，China： Association for Computational Linguistics： 11288-11309 ［DOI： 10.18653/v1/2025.findings-emnlp.604］
RT-H： action hierarchies using language
BelkhaleS，DingT L，XiaoT，SermanetP，VuongQ，TompsonJ，et al. 2024. RT-H： action hierarchies using language//Robotics： Science and Systems XX. Delft，the Netherlands： ［s.n.］
RT-1： robotics transformer for real-world control at scale
BrohanA，BrownN，CarbajalJ，ChebotarY，DabisJ，FinnC，et al. 2023. RT-1： robotics transformer for real-world control at scale//Robotics： Science and Systems XIX. Daegu，Korea（South）： ［s.n.］
OpenGVL-benchmarking visual temporal progress for data curation
BudzianowskiP，WiśniosE，TyrolskiM，GóralG，KulakovI，PetrenkoV，et al. 2025. OpenGVL-benchmarking visual temporal progress for data curation ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.17321.pdf
CompareBench： a benchmark for visual comparison reasoning in vision-language models
CaiJ，YangK N，FuL，DingJ M，LiJ L，SunH M，et al. 2025. CompareBench： a benchmark for visual comparison reasoning in vision-language models ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.22737.pdf
Matterport3D： learning from RGB-D data in indoor environments
ChangA，DaiA，FunkhouserT，HalberM，NiessnerM，SavvaM，et al. 2017. Matterport3D： learning from RGB-D data in indoor environments//Proceedings of 2017 International Conference on 3D Vision （3DV）. Qingdao，China： IEEE： 667-676 ［DOI： 10.1109/3DV.2017.00081］
ScanRefer： 3D object localization in RGB-D scans using natural language
ChenD Z，ChangA X and NießnerM. 2020. ScanRefer： 3D object localization in RGB-D scans using natural language//Proceedings of the 16th European Conference on Computer Vision — ECCV 2020. Glasgow，UK： Springer： 202-221 ［DOI： 10.1007/978-3-030-58565-5_13］
Scan2Cap： context-aware dense captioning in RGB-D scans
ChenD Z，GholamiA，Nießner M and ChangA X. 2021. Scan2Cap： context-aware dense captioning in RGB-D scans//Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville，USA： IEEE： 3192-3202 ［DOI： 10.1109/CVPR46437.2021.00321］
SafeMind： benchmarking and mitigating safety risks in embodied LLM agents
ChenR L，SunY Q，WangJ H，LyuM Y，ZhangQ and ZengY. 2025a. SafeMind： benchmarking and mitigating safety risks in embodied LLM agents ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.25885.pdf
RoboTwin
ChenT X，ChenZ X，ChenB J，CaiZ J，LiuY B，LiZ X，et al. 2025b. RoboTwin 2.0： a scalable data generator and benchmark with strong domain randomization for robust bimanual robotic manipulation ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2506.18088.pdf
MineAnyBuild： benchmarking spatial planning for open-world ai agents
ChenZ，GholamiA，Nießner M and ChangA X. 2025c. MineAnyBuild： benchmarking spatial planning for open-world ai agents［EB/OL］.［2025-11-05］.
https://arxiv.org/pdf/2505.20148.pdf
OmniVTLA： vision-tactile-language-action model with semantic-aligned tactile sensing
ChengZ X，ZhangY Q，ZhangW K，LiH Y，WangK Y，SongL，et al. 2025. OmniVTLA： vision-tactile-language-action model with semantic-aligned tactile sensing ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2508.08706.pdf
Can-Do！ A dataset and neuro-symbolic grounded framework for embodied planning with large multimodal models
ChiaY K，SunQ，BingL D and PoriaS. 2024. Can-Do！ A dataset and neuro-symbolic grounded framework for embodied planning with large multimodal models ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2409.14277.pdf
GraspVLA： a grasping foundation model pre-trained on billion-scale synthetic action data
DengS L，YanM，WeiS L，MaH X，YangY X，ChenJ Y，et al. 2025. GraspVLA： a grasping foundation model pre-trained on billion-scale synthetic action data ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2505.03233.pdf
Task-driven SLAM benchmarking for robot navigation
DuY W，FengS Y，CortC G and VelaP A. 2025. Task-driven SLAM benchmarking for robot navigation//Proceedings of 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Hangzhou，China： IEEE： 7147-7153 ［DOI： 10.1109/IROS60139.2025.11246415］
A survey of embodied AI： from simulators to research tasks
DuanJ F，YuS，TanH L，ZhuH Y and TanC. 2022. A survey of embodied AI： from simulators to research tasks. IEEE Transactions on Emerging Topics in Computational Intelligence，6（2）： 230-244 ［DOI： 10.1109/TETCI.2022.3141105］
RH20T： a comprehensive robotic dataset for learning diverse skills in one-shot
FangH S，FangH J，TangZ Y，LiuJ R，WangC X，WangJ B，et al. 2024. RH20T： a comprehensive robotic dataset for learning diverse skills in one-shot//Proceedings of 2024 IEEE International Conference on Robotics and Automation （ICRA）. Yokohama，Japan： IEEE： 653-660 ［DOI： 10.1109/ICRA57147.2024.10611615］
ActionNet： a dataset for dexterous bimanual manipulation
Fourier ActionNet Team and Mu Y. 2025. ActionNet： a dataset for dexterous bimanual manipulation ［EB/OL］. ［2025-11-05］.
https://action-net.org/
HomeSafeBench： a benchmark for embodied vision-language models in free-exploration home safety inspection
GaoS Y，YaoJ S，WenH Y，GuoY H，LiuZ M and HuangH Y. 2025. HomeSafeBench： a benchmark for embodied vision-language models in free-exploration home safety inspection［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.23690.pdf
Are we ready for autonomous driving？ The KITTI vision benchmark suite
GeigerA，LenzP and UrtasunR. 2012. Are we ready for autonomous driving？ The KITTI vision benchmark suite//Proceedings of 2012 IEEE Conference on Computer Vision and Pattern Recognition. Providence，USA： IEEE： 3354-3361 ［DOI： 10.1109/CVPR.2012.6248074］
RoboVerse： towards a unified platform，dataset and benchmark for scalable and generalizable robot learning
GengH R，WangF S，WeiS L，LiY Y，WangB J，AnB S，et al. 2025. RoboVerse： towards a unified platform，dataset and benchmark for scalable and generalizable robot learning ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2504.18904.pdf
USIM and U0： a vision-language-action dataset and model for general underwater robots
GuJ W，WuZ H，SiP X，QiuS，FengY K，SunL Y，et al. 2025. USIM and U0： a vision-language-action dataset and model for general underwater robots ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2510.07869.pdf
RT-Trajectory： robotic task generalization via hindsight trajectory sketches
GuJ Y，KirmaniS，WohlhartP，LuY，Gonzalez ArenasM，RaoK，et al. 2024. RT-Trajectory： robotic task generalization via hindsight trajectory sketches//Proceedings of the 12th International Conference on Learning Representations. Vienna，Austria： OpenReview.net
a unified benchmark for generalizable manipulation skills
GuJ Y，XiangF B，LiX L，LingZ，LiuX Q，MuT Z，et al. 2023. ManiSkill2： a unified benchmark for generalizable manipulation skills//Proceedings of the 11th International Conference on Learning Representations. Kigali，Rwanda： OpenReview.net
BEDI： a comprehensive benchmark for evaluating embodied agents on UAVs
GuoM N，WuM W，HeJ R，LiS X，LiH F and TaoC. 2025. BEDI： a comprehensive benchmark for evaluating embodied agents on UAVs［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2505.18229.pdf
RoboCerebra： a large-scale benchmark for long-horizon robotic manipulation evaluation
HanS H，QiuB X，LiaoY，HuangS Y，GaoC，YanS C，et al. 2025. RoboCerebra： a large-scale benchmark for long-horizon robotic manipulation evaluation ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2506.06677.pdf
A framework for benchmarking and aligning task-planning safety in LLM-based embodied agents
HuangY T，DingL L，TangZ P，WangT F，LinX R，ZhangW Y，et al. 2025. A framework for benchmarking and aligning task-planning safety in LLM-based embodied agents ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2504.14650.pdf
InternData： a large-scale multimodal synthetic and real hybrid dataset for embodied intelligence
RLBench： the robot learning benchmark and learning environment
JamesS，MaZ C，ArrojoD R and DavisonA J. 2020. RLBench： the robot learning benchmark and learning environment. IEEE Robotics and Automation Letters，5（2）： 3019-3026
Beyond the destination： a novel benchmark for exploration-aware embodied question answering
JiangK X，LiuY，ChenW X，LuoJ Z，ChenZ L，PanL，et al. 2025. Beyond the destination： a novel benchmark for exploration-aware embodied question answering ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2503.11117.pdf
Reinforcement learning： a survey
KaelblingL P，LittmanM L and MooreA W. 1996. Reinforcement learning： a survey. Journal of Artificial Intelligence Research，4（1）： 237-285
Arena-Rosnav 2.0： a development and benchmarking platform for robot navigation in highly dynamic environments
KästnerL，CarstensR，ZengH J，KmiecikJ，BhuiyanT，KhorsandhiN，et al. 2023. Arena-Rosnav 2.0： a development and benchmarking platform for robot navigation in highly dynamic environments//Proceedings of 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Detroit，USA： IEEE： 11257-11264 ［DOI： 10.1109/IROS55552.2023.10342152］
GOAT-Bench： a benchmark for multi-modal lifelong navigation
KhannaM，RamrakhyaR，ChhablaniG，YenamandraS，GervetT，ChangM，et al. 2024. GOAT-Bench： a benchmark for multi-modal lifelong navigation//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle，USA： IEEE： 16373-16383 ［DOI： 10.1109/CVPR52733.2024.01549］
DROID： a large-scale in-the-wild robot manipulation dataset
KhazatskyA，PertschK，NairS，BalakrishnaA，DasariS，KaramchetiS，et al. 2024. DROID： a large-scale in-the-wild robot manipulation dataset//Robotics： Science and Systems XX. Delft，the Netherlands： ［s.n.］
Building machines that learn and think like people
LakeB M，UllmanT D，TenenbaumJ B and GershmanS J. 2017. Building machines that learn and think like people. Behavioral and Brain Sciences，40： #253 ［DOI： 10.1017/S0140525X16001837］
Deep learning
LeCunY，BengioY and HintonG. 2015. Deep learning. Nature，521（7553）： 436-444 ［DOI： 10.1038/nature14539］
2025
LeeJ，MiyanishiT，KuritaS，SakamotoK，AzumaD，MatsuoY，et al. 2025. CityNav： a large-scale dataset for real-world aerial navigation//Proceedings of 2025 IEEE/CVF International Conference on Computer Vision. 5912-5922
Image quality assessment for embodied AI
LiC Y，XiaoJ H，ZhangJ B，WenF R，ZhangZ C，TianY，et al. 2025a. Image quality assessment for embodied AI ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2505.16815.pdf
R-Bench： are your large multimodal model robust to real-world corruptions？
LiC Y，ZhangJ B，ZhangZ C，WuH N，TianY，SunW，et al. 2025b. R-Bench： are your large multimodal model robust to real-world corruptions？ IEEE Journal of Selected Topics in Signal Processing，19（7）： 1349-1361 ［DOI： 10.1109/JSTSP.2025.3558652］
Evaluating real-world robot manipulation policies in simulation
LiX，HsuK，GuJ，PertschK，MeesO，WalkeH R，et al. 2025c. Evaluating real-world robot manipulation policies in simulation//Proceedings of the 8th Conference on Robot Learning. Munich，Germany： PMLR： 3705-3728
Large model empowered embodied AI： a survey on decision-making and embodied learning
LiangW L，ZhouR，MaY，ZhangB，LiS L，LiaoY J，et al. 2025. Large model empowered embodied AI： a survey on decision-making and embodied learning ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2508.10399.pdf
LIBERO： benchmarking knowledge transfer for lifelong robot learning
LiuB，ZhuY F，GaoC K，FengY H，LiuQ，ZhuY K，et al. 2023. LIBERO： benchmarking knowledge transfer for lifelong robot learning//Proceedings of the 37th International Conference on Neural Information Processing Systems. New Orleans，USA： Curran Associates Inc.： #1939
RoboView-Bias： benchmarking visual bias in embodied agents for robotic manipulation
Na Vid-4D： unleashing spatial intelligence in egocentric RGB-D videos for vision-and-language navigation
LiuH R，WanW K，YuX Q，LiM H，ZhangJ Z，ZhaoB，et al. 2025b. Na Vid-4D： unleashing spatial intelligence in egocentric RGB-D videos for vision-and-language navigation//Proceedings of 2025 IEEE International Conference on Robotics and Automation. Atlanta，USA： IEEE： 10607-10615 ［DOI： 10.1109/ICRA55743.2025.11128467］
EgoTraj-Bench： towards robust trajectory prediction under ego-view noisy observations
LiuJ Y，ZhouJ M，YeK，LinK Y，WangA and LiangJ W. 2025c. EgoTraj-Bench： towards robust trajectory prediction under ego-view noisy observations ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2510.00405.pdf
IS-Bench： evaluating interactive safety of VLM-driven embodied agents in daily household tasks
FMB： a functional manipulation benchmark for generalizable robotic learning
LuoJ L，XuC，LiuF C，TanL M，LinZ P，WuJ，et al. 2025. FMB： a functional manipulation benchmark for generalizable robotic learning. The International Journal of Robotics Research，44（4）： 592-606 ［DOI： 10.1177/02783649241276017］
PhyBlock： a progressive benchmark for physical understanding and planning via 3D block assembly
MaL，WenJ J，LinM，XuR T，LiangX W，LinB Q，et al. 2025. PhyBlock： a progressive benchmark for physical understanding and planning via 3D block assembly ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2506.08708.pdf
： situated question answering in 3D scenes
MaX J，YongS L，ZhengZ L，LiQ，LiangY T，ZhuS C，et al. 2023. SQA3D： situated question answering in 3D scenes//Proceedings of the 11th International Conference on Learning Representations. Kigali，Rwanda： OpenReview.net
DRISHTIKON： a multimodal multilingual benchmark for testing language models’ understanding on Indian culture
MajiA，KumarR，GhoshA，Anushka，ShahN，BorahA，et al. 2025. DRISHTIKON： a multimodal multilingual benchmark for testing language models’ understanding on Indian culture//Proceedings of 2025 Conference on Empirical Methods in Natural Language Processing. Suzhou，China： Association for Computational Linguistics： 1289-1313 ［DOI： 10.18653/v1/2025.emnlp-main.68］
OpenEQA： embodied question answering in the era of foundation models
MajumdarA，AjayA，ZhangX H，PuttaP，YenamandraS，HenaffM，et al. 2024. OpenEQA： embodied question answering in the era of foundation models//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Seattle，USA： IEEE： 16488-16498 ［DOI： 10.1109/CVPR52733.2024.01560］
Isaac Gym： high performance GPU-based physics simulation for robot learning
MakoviychukV，WawrzyniakL，GuoY R，LuM，StoreyK，Macklin，M，et al. 2021. Isaac Gym： high performance GPU-based physics simulation for robot learning//Proceedings of the 35th International Conference on Neural Information Processing Systems. ［s.l.］： Curran Associates Inc.
CALVIN： a benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks
MeesO，HermannL，Rosete-BeasE and BurgardW. 2022. CALVIN： a benchmark for language-conditioned policy learning for long-horizon robot manipulation tasks. IEEE Robotics and Automation Letters，7（3）： 7327-7334 ［DOI： 10.1109/LRA.2022.3180108］
Agent-RewardBench： towards a unified benchmark for reward modeling across perception，planning，and safety in real-world multimodal agents
MenT Y，JinZ R，CaoP F，ChenY B，LiuK and ZhaoJ. 2025. Agent-RewardBench： towards a unified benchmark for reward modeling across perception，planning，and safety in real-world multimodal agents//Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics （Volume 1： Long Papers）. Vienna，Austria： Association for Computational Linguistics： 17521-17541 ［DOI： 10.18653/v1/2025.acl-long.857］
Research on virtual embodied interaction technology for VR physics experiments
MengQ F，ZhangY R，ZhangH W，HuX Y and LuoY H. 2025. Research on virtual embodied interaction technology for VR physics experiments. Journal of Image and Graphics，1-12
Spatial reasoning in foundation models： benchmarking object-centric spatial understanding
MirjaliliV，GiahiR，KolliparaS，KekudaA，YaoK H，ZhaoK，et al. 2025. Spatial reasoning in foundation models： benchmarking object-centric spatial understanding ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.21922.pdf
RoboCasa： large-scale simulation of everyday tasks for generalist robots
NasirianyS，MaddukuriA，ZhangL C，ParikhA，LoA，JoshiA，et al. 2024. RoboCasa： large-scale simulation of everyday tasks for generalist robots ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2406.02523.pdf
Embodied arena： a comprehensive，unified，and evolving evaluation platform for embodied AI
NiF，ZhangM，LiP Y，YuanY F，ZhangL F，LiuY C，et al. 2025. Embodied arena： a comprehensive，unified，and evolving evaluation platform for embodied AI ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.15273.pdf
Physical backdoor attack can jeopardize driving with vision-large-language models
NiZ Y，YeR，WeiY X，XiangZ，WangY F and ChenS H. 2024. Physical backdoor attack can jeopardize driving with vision-large-language models ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2404.12916.pdf
Open X-embodiment： robotic learning datasets and RT-X models
O’NeillA，RehmanA，GuptaA，MaddukuriA，GuptaA，PadalkarA，et al. 2023. Open X-embodiment： robotic learning datasets and RT-X models ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2310.08864.pdf
O’NeillA，RehmanA，MaddukuriA，GuptaA，PadalkarA，LeeA，et al. 2024. Open X-embodiment： robotic learning datasets and RT-X models： open X-embodiment collaboration0//Proceedings of 2024 IEEE International Conference on Robotics and Automation. Yokohama，Japan： IEEE： 6892-6903 ［DOI： 10.1109/ICRA57147.2024.10611477］
How the Body Shapes the Way We Think： A New View of Intelligence. Cambridge，USA： The MIT Press
PfeiferR and BongardJ. 2006. How the Body Shapes the Way We Think： A New View of Intelligence. Cambridge，USA： The MIT Press
VLN-R1： vision-language navigation via reinforcement fine-tuning
QiZ Y，ZhangZ X，YuY Z，WangJ Q and ZhaoH S. 2025. VLN-R1： vision-language navigation via reinforcement fine-tuning ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2506.17221.pdf
Jailbreaking LLM-controlled robots
RobeyA，RavichandranZ，KumarV，HassaniH and PappasG J. 2025. Jailbreaking LLM-controlled robots//Proceedings of 2025 IEEE International Conference on Robotics and Automation （ICRA）. Atlanta，USA： IEEE： 11948-11956 ［DOI： 10.1109/ICRA55743.2025.11128119］
RoboVQA： multimodal long-horizon reasoning for robotics
SermanetP，DingT L，ZhaoJ，XiaF，DwibediD，GopalakrishnanK，et al. 2024. RoboVQA： multimodal long-horizon reasoning for robotics//Proceedings of 2024 IEEE International Conference on Robotics and Automation （ICRA）. Yokohama，Japan： IEEE： 645-652 ［DOI： 10.1109/ICRA57147.2024.10610216］
A comprehensive survey on embodied intelligence： advancements，challenges，and future perspectives
SunF C，ChenR F，JiT Y，LuoY，ZhouH D and LiuH P. 2024. A comprehensive survey on embodied intelligence： advancements，challenges，and future perspectives. CAAI Artificial Intelligence Research，3： #9150042 ［DOI： 10.26599/AIR.2024.9150042］
ManiSkill3： GPU parallelized robotics simulation and rendering for generalizable embodied AI
TaoS，XiangF B，ShuklaA，QinY Z，HinrichsenX，YuanX D，et al. 2024. ManiSkill3： GPU parallelized robotics simulation and rendering for generalizable embodied AI ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2410.00425.pdf
Unbiased look at dataset bias
TorralbaA and EfrosA A. 2011. Unbiased look at dataset bias//Proceedings of the CVPR 2011. Colorado Springs，USA： IEEE： 1521-1528 ［DOI： 10.1109/CVPR.2011.5995347］
Continuous-time estimation for dynamic obstacle tracking
UshaniA K，Carlevaris-BiancoN，CunninghamA G，GalceranE and EusticeR M. 2015. Continuous-time estimation for dynamic obstacle tracking//Proceedings of 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Hamburg，Germany： IEEE： 1137-1143 ［DOI： 10.1109/IROS.2015.7353513］
GRUtopia： dream general robots in a city at scale
WangH Q，ChenJ H，HuangW S，BenQ W，WangT，MiB Y，et al. 2024a. GRUtopia： dream general robots in a city at scale ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2407.10943.pdf
Scaling proprioceptive-visual learning with heterogeneous pre-trained transformers
WangL R，ChenX L，ZhaoJ L and HeK M. 2024b. Scaling proprioceptive-visual learning with heterogeneous pre-trained transformers//Proceedings of the 38th International Conference on Neural Information Processing Systems. Vancouver，Canada： Curran Associates Inc.： #3952
VLATest： testing and evaluating vision-language-action models for robotic manipulation. Proceedings of the ACM on Software Engineering，2
WangZ J，ZhouZ H，SongJ Y，HuangY H，ShuZ and MaL. 2025. VLATest： testing and evaluating vision-language-action models for robotic manipulation. Proceedings of the ACM on Software Engineering，2（FSE）： 1615-1638 ［DOI： 10.1145/3729343］
RoboMIND： benchmark on multi-embodiment intelligence normative data for robot manipulation
WuK，HouC K，LiuJ M，CheZ P，JuX Z，YangZ Q，et al. 2024. RoboMIND： benchmark on multi-embodiment intelligence normative data for robot manipulation ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2412.13877.pdf
ViC-bench： benchmarking visual-interleaved chain-of-thought capability in MLLMs with free-style intermediate state representations
WuX C，LiuJ X，HuangD L，WangY F，ShiY Y，ChenK D，et al. 2025. ViC-bench： benchmarking visual-interleaved chain-of-thought capability in MLLMs with free-style intermediate state representations ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2505.14404.pdf
Data assessment for embodied intelligence
XiaoJ H，YanB W，ZhangJ B，WangJ，LiC Y，ChengZ X，et al. 2025a. Data assessment for embodied intelligence ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2508.06553.pdf
Static and plugged： make embodied evaluation simple
XiaoJ H，ZhangJ B，YanB W，GuoS Y，YeT R，ZhangK W，et al. 2025b. Static and plugged： make embodied evaluation simple［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2508.06553.pdf
Verti-bench： a general and scalable off-road mobility benchmark for vertically challenging terrain
XuT，PanC，RaoM B，DatarA，PokhrelA，LuY，et al，2025. Verti-bench： a general and scalable off-road mobility benchmark for vertically challenging terrain ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2502.11426.pdf
Point-it-out： benchmarking embodied reasoning for vision language models in multi-stage visual grounding
XueH，GeY，ZengY，LiZ，LiuM Y，ChenY，et al. 2025. Point-it-out： benchmarking embodied reasoning for vision language models in multi-stage visual grounding ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2509.25794.pdf
RoboTron-Mani： all-in-one multimodal large model for robotic manipulation
YanF，LiuF F，ZhengL M，ZhongY F，HuangY Y，GuanZ C，et al. 2024. RoboTron-Mani： all-in-one multimodal large model for robotic manipulation ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2412.07215.pdf
Thinking in space： how multimodal large language models see，remember，and recall spaces
YangJ，YangS，GuptaA W，HanR，Fei-FeiL and XieS. 2025. Thinking in space： how multimodal large language models see，remember，and recall spaces ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2412.14171.pdf
SafeAgentBench： a benchmark for safe task planning of embodied LLM agents
YinS，PangX H，DingY Z，ChenM L，BiY T，XiongY C，et al. 2024. SafeAgentBench： a benchmark for safe task planning of embodied LLM agents ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2412.13178.pdf
AGENTSAFE： benchmarking the safety of embodied agents on hazardous instructions
YingZ H，WangL，XiaoY S，WangJ K，MaY Q，GuoJ Y，et al. 2025. AGENTSAFE： benchmarking the safety of embodied agents on hazardous instructions ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2506.14697.pdf
Meta-world： a benchmark and evaluation for multi-task and meta reinforcement learning
YuT H，QuillenD，HeZ P，JulianR，HausmanK，FinnC，et al. 2019. Meta-world： a benchmark and evaluation for multi-task and meta reinforcement learning//Proceedings of the 3rd Annual Conference on Robot Learning. Osaka，Japan： PMLR： 1094-1100
From seeing to doing： bridging reasoning and decision for robotic manipulation
YuanY F，CuiH Q，ChenY B，DongZ B，NiF，KouL X，et al. 2025. From seeing to doing： bridging reasoning and decision for robotic manipulation ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2505.08548.pdf
BadRobot： jailbreaking LLM-based embodied AI in the physical world
ZhangH，ZhuC，WangX，ZhouZ，Hu S and ZhangL Y，2024d. BadRobot： jailbreaking LLM-based embodied AI in the physical world ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2407.20242.pdf
Embodied image quality assessment for robotic intelligence
Uni-NaVid： a video-based vision-language-action model for unifying embodied navigation tasks
ZhangJ Z，WangK Y，WangS A，LiM H，LiuH R，WeiS L，et al. 2024b. Uni-NaVid： a video-based vision-language-action model for unifying embodied navigation tasks ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2412.06224.pdf
NaVid： video-based VLM plans the next step for vision-and-language navigation
ZhangJ Z，WangK Y，XuR T，ZhouG Z，HongY C，FangX M，et al. 2024c. NaVid： video-based VLM plans the next step for vision-and-language navigation//Robotics： Science and Systems XX. Delft，the Netherlands： ［s.n.］
ET-plan-bench： embodied task-level planning benchmark towards spatial-temporal cognition with foundation models
ZhangL F，WangY N，GuH J，HamidizadehA，ZhangZ G，LiuY C，et al. 2025a. ET-plan-bench： embodied task-level planning benchmark towards spatial-temporal cognition with foundation models//Proceedings of 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）. Hangzhou，China： IEEE： 21566-21573 ［DOI： 10.1109/IROS60139.2025.11246658］
VLABench： a large-scale benchmark for language-conditioned robotics manipulation with long-horizon reasoning tasks
ZhangS D，XuZ，LiuP J，YuX P，LiY，GaoQ H，et al. 2025b. VLABench： a large-scale benchmark for language-conditioned robotics manipulation with long-horizon reasoning tasks//Proceedings of 2025 IEEE/CVF International Conference on Computer Vision. ［s.l.］： ［s.n.］： 11142-11152
EmbodiedVSR： dynamic scene graph-guided chain-of-thought reasoning for visual spatial tasks
ZhangY，ZhangQ，JuX Z，LiuZ Y，MaoJ L，SunJ K，et al. 2025c. EmbodiedVSR： dynamic scene graph-guided chain-of-thought reasoning for visual spatial tasks ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2503.11089.pdf
AIBench： towards trustworthy evaluation under the 45°law
ZhangZ C，WangJ Y，GuoY J，WenF R，ChenZ J，WangH Q，et al. 2026. AIBench： towards trustworthy evaluation under the 45°law. Displays，91： #103255 ［DOI： 10.1016/j.displa.2025.103255］
Large multimodal models evaluation： a survey
ZhengZ C，WangJ Y，WenF R，GuoY J，ZhaoX Y，FangX Y，et al. 2025d. Large multimodal models evaluation： a survey. Science China Information Sciences，68（12）： #221301 ［DOI： 10.1007/s11432-025-4676-4］
RoboCAS： a benchmark for robotic manipulation in complex object arrangement scenarios
ZhengL M，YanF，LiuF F，FengC J，KangZ L and MaL. 2024. RoboCAS： a benchmark for robotic manipulation in complex object arrangement scenarios ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2407.06951.pdf
ZhongN H，CaroS，IskandarA，RameshM and SmithS L. 2025. Bench-NPIN： benchmarking non-prehensile interactive navigation［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2505.12084.pdf
PhysVLM： enabling visual language models to understand robotic physical reachability
ZhouW J，TaoM L，ZhaoC Y，GuoH Y，DongH H，TangM，et al. 2025a. PhysVLM： enabling visual language models to understand robotic physical reachability//Proceedings of 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR）. Nashville，USA： 6940-6949 ［DOI： 10.1109/CVPR52734.2025.00651］
AutoEval： autonomous evaluation of generalist robot manipulation policies in the real world
ZhouZ Y，AtreyaP，TanY L，PertschK and LevineS. 2025b. AutoEval： autonomous evaluation of generalist robot manipulation policies in the real world ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2503.24278.pdf
EARBench： towards evaluating physical risk awareness for task planning of foundation model-based embodied AI agents
ZhuZ H，WuB Z，ZhangZ Y，HanL，LiuQ S and WuB Y. 2024. EARBench： towards evaluating physical risk awareness for task planning of foundation model-based embodied AI agents ［EB/OL］. ［2025-11-05］.
https://arxiv.org/pdf/2408.04449.pdf
RT-2： vision-language-action models transfer web knowledge to robotic control
ZitkovichB，YuT H，XuS C，XuP，XiaoT，XiaF，et al. 2023. RT-2： vision-language-action models transfer web knowledge to robotic control//Proceedings of 2023 Conference on Robot Learning. Atlanta，USA： PMLR： 2165-2183
第 1 页 / 共 1 页
2026年第31卷第8期
PDF下载
29
15
引用本文
BibTeX
文章信息
doi: 10.11834/jig.250550
接收时间：2025-11-05
首发时间：2026-10-30
出版时间：2026-08-16
补充材料
相关文章
文章信息
作者
出版历史
收稿日期：2025-11-05
修回日期：2026-01-29
基金
National Natural Science Foundation of China（62225112，625B2118）；Singapore MoE Foundation（MOE-T2EP20123-0006）
PDF下载
全屏
29
15
关闭全屏
北京市海淀区学院南路86号
010-62662699（运营及技术支持）
qkjq@cast.org.cn
