A Tactile Sensor Array with a Monolithically Integrated Neural Network for Edge Computation

Described presently is a monolithically integrated "intelligent" sensor system consisting of a piezoelectric tactile sensor array and an artificial neural network (ANN). Low-temperature processed, dual-gate thin-film transistors (TFTs) based on semiconducting indium-tin-zinc oxide are deployed in the former for signal acquisition and in the latter for executing the "multiply-accumulate" operation. With the ANN appropriately trained, the system is capable of both "imaging" and "recognizing" a distributed force load applied on the sensor array using 3D-printed stamps. The proposed integration scheme can be deployed to construct intelligent sensor systems for applications other than tactile sensing.