AI-powered electronic skin: from smart design to cognitive interaction

Robotic electronic skin (e-skin) is inspired by human skin and endows robots with tactile perception, temperature detection, and environmental interaction capabilities. However, its development is hampered by prolonged design cycles, limited signal enhancement, and weak cognitive abilities. Given that the convergence of artificial intelligence (AI) with e-skin is fundamentally transforming this landscape, the present review highlights the pivotal contributions of AI across the entire development spectrum of robotic e-skin, including design optimization, signal processing, and cognitive enhancement. AI-driven design paradigms dramatically shorten development time and enable the discovery of optimal sensor materials and structures. In signal processing, AI algorithms notably improve the ability to decouple complex sensory data, enabling robust, multimodal, super-resolution sensing. AI endows e-skin with advanced cognitive capabilities, allowing it to interpret intricate tactile information and intelligently respond to external environments. By underscoring the potential of AI throughout the entire development pipeline, this review aims to drive the creation of e-skin with minimal hardware and maximal cognition and thus achieve revolutionary breakthroughs in cutting-edge fields such as human-robot interactions, precise robot control, and soft robotics for environmental exploration.