Intelligent Recognition and Control System for Flexible Robotic Hands Based on Multi-channel Tactile Sensing

With the development of intelligent manufacturing and robotic technology, flexible robotic hands have demonstrated substantial application potential in fields such as industrial automation and medical devices due to their flexibility and adaptability. However, traditional visual recognition methods have limitations in acquiring mechanical properties of objects, such as hardness and texture, which makes it difficult to achieve accurate recognition under complex environments. To address these limitations, this paper proposes a multi-channel tactile sensing system based on the STM32F1 microcontroller, aiming to endow the flexible robotic hand with efficient object recognition and grasp control capabilities. The system integrates multiple thin-film pressure sensors, with a Convolutional Neural Network (CNN) employed to process the pressure data in multiple channels, extracting tactile features layer by layer and analyzing the dynamic mechanical information during the grasping process. Experimental results indicate that the designed system achieves an accuracy of 97.9% in the classification tasks involving objects with various hardness levels, verifying the robustness and accuracy of the system in identifying diverse objects.