Abstract:A smart sensor architecture combining an array of sensing elements with an overlapping array of computing and memory elements thus emulates an innervated peripheral sensing system (IPSS) capable of local and autonomous neuromorphic in‐sensor data pre‐processing is presented. Compatibility of the proposed architecture with functionally distinct elements for sensing, memory, and computing removes the restrictive demand for a single element simultaneously capable of all, thus making this architecture more generally applicable to a wider range of sensors and usage scenarios. An array of sensors generating a collection of correlated signals can benefit from integration with a "smart" system for autonomous inferencing. Mimicking their biological counterparts, smart sensor systems shall possess the capabilities of sensing, memory, and neuromorphic computation. However, state‐of‐the‐art biomimetic systems either do not employ a full set of devices to cover the complete range of capabilities or incorporate devices that are capable of all but appropriate only for a limited range of sensing applications. Presently proposed is a smart sensor architecture that combines an array of sensing elements with an overlapping array of computing and memory elements, thus emulating an innervated peripheral sensing system (IPSS) capable of local and autonomous neuromorphic in‐sensor data pre‐processing. Compatibility of the proposed architecture with functionally distinct elements for sensing, memory, and computing removes the restrictive demand for a single element simultaneously capable of all, thus making this architecture more generally applicable to a wider range of sensors and usage scenarios. An artificial synapse as a computing element is implemented using dual‐gate (DG) thin‐film transistors (TFTs) and the low‐leakage current of transistors based on metal‐oxide semiconductors allows the deployment of capacitors as memory elements. The outputs of the IPSS are passed on to an adjacent artificial neural network (ANN) for near‐sensor inferencing. Monolithic integration of the IPSS and the ANN is made possible by the deployment of the same memory and computing elements in their construction. A smart tactile sensing system based on the proposed architecture is constructed and characterized. The functionality of the system is demonstrated by its application to the classification of a set of tactile images of 3‐dimensionally printed alphabet stamps.

In-sensor Computing ANN Capacitive Sensors

A Reconfigurable Convolution-in-Pixel CMOS Image Sensor Architecture

Computing Resistance-Style Image Sensors for Artificial Neural Networks

An intelligent pressure sensor using neural networks

C3PU: Cross-Coupling Capacitor Processing Unit Using Analog-Mixed Signal In-Memory Computing for AI Inference

In‐Sensor Computing‐Based Smart Sensing Architecture Implemented Using a Dual‐Gate Metal‐Oxide Thin‐Film Transistor Technology

An Analog-Memoryless Near Sensor Computing Architecture for Always-On Intelligent Perception Applications

MACSen: A Processing-In-Sensor Architecture Integrating MAC Operations Into Image Sensor for Ultra-Low-Power BNN-Based Intelligent Visual Perception

Multiframe-integrated, In-Sensor Computing Using Persistent Photoconductivity

In-sensor nonlinear convolutional processing based on hybrid MTJ/CMOS arrays

NAS-Cap: Deep-Learning Driven 3-D Capacitance Extraction with Neural Architecture Search and Data Augmentation

An in-memory computing architecture based on two-dimensional semiconductors for multiply-accumulate operations

An Energy-Efficient Flexible Capacitive Pressure Sensing System.

Design of a Capacitive Tactile Sensor Array System for Human–Computer Interaction

In-sensor neural network for high energy efficiency analog-to-information conversion

A Digital-Analog Hybrid System-on-Chip for Capacitive Sensor Measurement and Control.

6.9 A 0.35V 0.367TOPS/W Image Sensor with 3-Layer Optical-Electronic Hybrid Convolutional Neural Network

In-situ artificial retina with all-in-one reconfigurable photomemristor networks

Low-Cost, Flexible Annular Interdigital Capacitive Sensor (faics) with Carbon Black-Pdms Sensitive Layer for Proximity and Pressure Sensing

An 11T1C Bit-Level-Sparsity-Aware Computing-in-Memory Macro with Adaptive Conversion Time and Computation Voltage

Spiking Neural Network Based on Memory Capacitors and Metal-Oxide Thin-Film Transistors