A Multisensory Tactile System for Robotic Hands to Recognize Objects

摘要

Abstract Multisensory tactile systems play an important role in enhancing robot intelligence. A competent robotic tactile system needs to be simple in structure and easily operated, especially with multiple sensations, and have good coordinate ability like human skin. A novel multisensory tactile system for humanoid robotic hands is proposed, allowing the hand to identify objects by grasping and manipulating them. Robotic multifunction sensors based on skin‐inspired thermosensation and structured with micro platinum ribbons partially covered with piezo‐thermic silver‐nanoparticle‐doped porous polydimethylsiloxane membrane simultaneously and independently detect contact pressure, local ambient temperature, and thermal conductivity and temperature of an object. Multiple tactile information which is obtained by the robotic sensors is fused comprehensively based on neural network classification to identify diverse objects in uncertain and dynamic gripping with an object recognition accuracy of 95%. This multisensory system enables robotic hand to better interact with its environment, enhances robotic intelligence, and makes various complex tasks feasible for robots, such as sorting material or rescuing from fire or other disasters.