Electronic skin: transforming robotic sensing

The development of neuro-inspired electronic skin (e-skin) is transforming the landscape of robotics, offering sophisticated touch-sensing capabilities that mimic human skin. This review delves into the advancements in hardware implementations, neural-like sensing, and data processing for e-skin, drawing parallels with the human peripheral nervous system (PNS).
Touch is a complex sensory modality that involves various receptors (mechanoreceptors, thermal, and pain receptors) embedded in the skin. These receptors enable humans to perceive and interact with their environment efficiently. Replicating this capability in robots requires innovative hardware and computational approaches. Neuro-inspired e-skin aims to achieve this by integrating distributed electronics and neuromorphic circuits originally developed for other sensory modalities.
The article discusses the hardware implementations of computational building blocks for e-skin, including sensors, neurons, and synapses, which can be integrated to replicate human-like tactile functionalities. For instance, flexible and printed electronics on soft substrates are essential for developing large-area e-skin. Ultrathin neuromorphic chips are highlighted for their potential to handle local computation, reducing data latency and power consumption.
Various approaches are explored to emulate the neural pathways involved in human tactile sensation and perception. This includes the use of oscillating circuits and neuron circuits to create neuromorphic sensors that output biologically plausible spiking signals. Additionally, phase-change devices and memristive devices are reviewed for their roles in mimicking neuron and synapse functionalities, enabling advanced tactile data processing.