A Programmable Electronic Skin with Event-Driven In-Sensor Touch Differential and Decision-Making

Abstract

High-precise, crosstalk-free tactile perception offers an intuitive way for informative human-machine interactions. However, the differentiation and labeling of touch position and strength require substantial computational space due to the cumbersome post-processing of parallel data. Herein, a programmable and robust electronic skin (PR e-skin) with event-driven in-sensor touch differential and perception, solving the inherent defects in the von Neumann framework is introduced. The PR e-skin realizes feature simplification and reduction of data transmission by integrating the computing framework into sensing terminals. Furthermore, the event-driven functional mode further greatly compresses untriggered redundant data. Benefiting from the minimal concise dataset, the PR e-skin can directly differentiate touch position and pressure with swift response time (<0.3 ms). Robust carbon functional film ensures long-term and stable implementation (>10 000 cycles) of the in-sensor computing architectural feature. In a designable, continuous position detection with an extensive pressure range (210 kPa), which is an improvement of 5.5 times, the PR e-skin can ultra-sensitive extract trajectory sliding or rapping actions. Moreover, combined with customized neural network, a dual-encryption recognition system is constructed based on slide action, reaching a high recognition accuracy of ≈98%, which reveals the great potential in intelligent interaction and security.