Synthetic Mechanoreceptors with Collocated Logic

We present a new class of synthetic mechanoreceptor exhibiting large changes in conductivity as a function of their shape. Concretely, such mechanoreceptors consist of a mechanical bistable structure showing shape-dependent electrical conductivity. The electromechanical response is designed such that on one state the unit shows large resistance, while on the second state the conductivity increases significantly. The bistable nature of such units and their ability to switch conductivity in response to external forcing allows our mechanoreceptors to serve as sensors with collocated input dependent memory. Specifically, the amplitude and frequency of the pressure/force inputs activating our mechanoreceptors change of shape (snap-through) result in distinct dynamical signatures, which are transduced by adding a voltage bias and reading current time histories. This allows to interpret specific external inputs into electrical output signatures that can be feed into an artificial neural network. We demonstrate this by subjecting a network of such mechanoreceptors to different force inputs resulting in recognizable electrical signatures which correlate to specific patterns of local states.