NRI: Robust and Low-Cost Smart Skin with Active Sensing Network for Enhancing Human-Robot Interaction

Stanford University, Stanford CA

Investigators

Abstract

Tactile sensing is ubiquitous in nature - arguably even more essential than vision. Most animals have thousands of cutaneous sensors over their bodies for touch, temperature, etc. But even the most sophisticated robots have relatively few tactile sensors and, after 30 years of research, tactile sensing lags behind computer vision. This project aims at the development of a novel artificial skin mimicking the human skin that can be fitted into any robotic hand providing information-rich "sense of touch." This technology leads to the development of extremely sensitive robotic skins with unprecedented tactile sensing capabilities. As such, this work enables a plethora of robotic applications where tactile sensing is of utmost importance, ranging from robotic caregivers to medical robotics and autonomous exploration. The methodology in this project may revolutionize the way future robots are designed enabling their broad applicability. This effort represents a major milestone in endowing robots with the sensory information required to carry out tasks in human-centered environments. The advent of microprocessors for touch sensing, spurred by the smart phone industry, has helped to address the wiring problem with local processing of information and communication. However, there remain the critical problems of fabricating a multi-functional artificial skin that can conformally cover arbitrary surfaces, diagnosing in real-time, the contact state, and gathering a large amount of data for high-resolution tactile sensing, while minimizing power consumption. Overall, this effort addresses tactile sensing from a system-level point of view. The approach involves the development of advanced manufacturing technologies from leveraging nonstandard CMOS/MEMS/NEMS fabrication processes to produce a low-cost and robust artificial skin outfitted with multi-modal micro-sensors. The tactile sense of touch is achieved via an innovative micro-contact sensing technique based on ultrasound waves generated from embedded sensors to identify local contact/slip conditions. Finally, the validation and performance evaluation is demonstrated through a series of graded tactile sensing experiments.