With the continuous development of tactile sensing technology，the application of piezoelectric materials in tactile sensors has garnered increasing attention. Currently，tactile sensors face challenges such as low recognition accuracy，insufficient response sensitivity，and poor stability in complex environments. To address these issues，research was conducted on utilizing the piezoelectric properties of polyvinylidene difluoride（PVDF）to convert external force signals into electrical signals for sensor design. Additionally，a microcontroller was utilized for real-time acquisition and storage of data collected by tactile sensors. At the same time，the improved Back Propagation（BP）neural network was combined withParticle Swarm Optimization（PSO）to enhance signal processing and recognition capabilities. The sensitivity and response accuracy of the sensor were significantly improved through the design of a PVDF multilayer structure. The results show that the classification performance（accuracy 98.54%，recall 98.13%，F1 value 97.42%）is significantly better than that of the comparison algorithm，and the highest recognition accuracy of the sensor for different roughness and hardness items reaches 95% and 96%，respectively，with a maximum root mean square error（RMSE）of only 0.032.In summary，the design of a PVDF piezoelectric film and single-chip tactile sensor based on the improved BP has effectively improved the response accuracy of tactile sensors under different tactile stimuli，with high perceptual sensitivity and stability. This further promotes the application and development of intelligent robots in precision operations and complex tasks.

improve BP neural network  /  PVDF piezoelectric film  /  single-chip microcomputer  /  tactile sensor  /  signal processing