Distributed and stretchable tactile sensing for dexterous robotic hands based on a crosslinked interpenetrating network
Tactile sensing for dexterous robotic hands is essential for achieving human-like precision in manipulation. However, current tactile sensors face challenges such as insufficient durability, limited coverage, and poor conformability to curved, jointed surfaces. This study presents a stretchable distributed tactile sensor array designed for dexterous robotic hands. The array comprises 18 sensing units distributed across the hand, incorporating quasi-homogeneous functional layers interconnected by crosslinked interpenetrating networks, and composite electrodes combining high conductivity with stretchability. This design yields a thin, soft, transparent, and stretchable sensor array that integrates seamlessly with a commercial dexterous hand. The sensor array exhibits high interlayer tensile strength, high sensitivity, low hysteresis, and excellent long-term reliability over 10,000 loading cycles. Experimental results demonstrate accurate detection of tactile force across the entire robotic hand during object grasping. Using convolutional neural network algorithms, the sensor array identifies different object types with 90.1% accuracy, with results displayed in real time on a digital twin interface. The proposed sensor array holds significant potential for embodied intelligence and robotics in adaptive grasping, safe manipulation, and remote teleoperation.