Multifunctional nanomaterials, systems, and algorithms for neuromorphic computing applications: Autonomous systems and wearable robotics

Abstract

Memristive devices are the preferred choice for neuromorphic computer architectures, with low-dimensional materials exhibiting unique functionality resembling biological neurons. The ability to adjust these properties presents significant opportunities for artificial neural networks. This review offers a critical investigation of emerging multi-functional (MF) neuromorphic devices enabled by zero-dimensional, one-dimensional, and two-dimensional materials, van der Waals heterojunctions, and their mechanisms. It highlights the multiple unique bio-inspired device responses that arises from quantum confinement, interfaces, and low-dimensional topology. The advancements, obstacles, and potential solutions for effective neuromorphic computing using low-dimensional MF neuromorphic systems are surveyed. This overview highlights the appealing attributes of neuromorphic computing for future computations and explores the potential for advancing neuromorphic algorithms based on low-dimensional MF systems. The development of low-dimensional MF neural networks for autonomous system applications is outlined. This review article investigates the integration of physical, physiological, and environmental data through low-dimensional MF neural networks, which is essential for wearable robotic applications. It also provides a prospective analysis of the opportunities and challenges associated with low-dimensional MF neuromorphic materials compared to conventional bulk electronic technologies.