Research Projects

Visual perception involves interpreting the surrounding environment, recognizing objects, and making decisions based on visual inputs. Developing artificial visual systems has become increasingly important due to the growing demand for autonomous applications, such as unmanned vehicles, robotics for defense, and remote surveillance systems. Currently, visual sensors or cameras rely on traditional von Neumann architectures, which face significant challenges, including high latency, excessive energy consumption, and limited parallelism due to the separation between storage and processing units. In contrast, biological synapses — composed of billions of neurons connected by trillions of synapses — excel in efficiently processing vast amounts of information, enabling rapid decision-making and parallel processing with minimal power consumption (around 20W). Artificial neuromorphic computing has emerged as a solution to address these challenges in data processing, storage, and parallel computing while maintaining low power consumption. Neuromorphic computers are designed to solve complex problems, recognize patterns, and make decisions instantaneously, opening up new possibilities for hardware-based AI. For India, advancing neuromorphic computation for visual synapses is particularly important to enhance self-reliance in AI technology, improve automated surveillance systems, and foster innovations in robotics for industries, healthcare and defense applications. The core innovation of this project lies in creating neuromorphic devices that truly mimic brain-inspired functionality by integrating sensing, memory, and processing into a single unit. To achieve this, we are leveraging the unique properties of 2D ferroelectric materials, specifically, light-induced polarization and bidirectional locking of dipoles. By optimizing these effects and examining dipole interactions, we aim to develop scalable, high-performance devices that can advance artificial neural networks (ANNs), machine learning (ML), and AI. In this project, we will investigate a some of 2D ferroelectric materials. Our focus will be on key performance factors such as material formation temperatures, the influence of contact metals on device performance, and the roles of these materials as both channel and gate components. We are specifically targeting three-terminal artificial synaptic devices, using these materials to simulate presynaptic signals through both optical and electrical inputs. Our goal is to demonstrate critical synaptic behaviors such as PSC and transitions from STM-LTM. Additionally, we aim to showcase advanced functionalities like plasticity, logical transformations, associative learning, image recognition, and multicolor pattern recognition. These hardware-based results will be tested through MNIST pattern recognition tasks, with potential applications in industry, healthcare, and defence sectors for contact-free remote surveillance.