Neuromorphic Sensing and Computing for Low-Power Systems

Brain inspired information processing in hardware and in software opens a possible path to real-time, low-energy computation, which is often wanted in on-the-edge computing or in wearable electronics. Today, first instances of such “neuromorphic computing systems” are available as customizable hardware units for in-field applications. For efficient use of such novel hardware, event-based or “spiking neuromorphic” algorithms are needed instead of traditional sequential (CPU) or parallel (GPU) computing. In this presentation, I will introduce the concepts of neuromorphic computing, present current and future neuromorphic hardware for sensing and computation, and discuss application examples of closed-loop sensing for actuated systems (robotics).