Low power sensor interface for artificial biomimetic skin
Tactile sensing is fundamental yet underestimated means for humans and robots to interact with the physical environment. Neuromorphic skin, which mimics the behavior and mechanical properties of the human tactile perception proves to be a more power-efficient approach. Previous research on neuromorphic skin, guided by mimicking simple receptive fields of monkeys, rely on a coarse sensor spatial resolution (>1mm) and a 1:1 sensor-to-neuron connection. Mimicking complex receptive fields in human skins, containing multiple highly-sensitive zones, however requires sub-mm sensor spatial resolution with many-to-one sensor-to-neuron connections. Quick estimates yield a required 10x density improvement of sensors over current state of the art. This increased sensor and consequently readout density requirement requires novel low-power, area-efficient readout while integrated on a flexible substrate.