Neuromorphic Agents

The fields of machine learning and computational neuroscience have made tremendous progresses in developing models and algorithms that mimic biophysical mechanisms that the nervous system uses to carry out computation. However, today’s artificial systems are still not able to compete with biological ones, in that they are typically power-hungry and operate with mainly serial and synchronous logic gates, with functions that are decoupled from their hardware implementation.

The project NeuroAgents aims at combining the recent advances in machine learning and neural computation with the latest achievements in neuromorphic technologies in order to build “neuromorphic agents”, i.e. autonomous electronic agents that can express cognitive behavior by sensing signals in real-time and adapting to unexpected changes in uncontrolled environments, while coping with noise and uncertainty.

The project team will follow an interdisciplinary approach to pursue a twofold objective: understanding the principles of cognition by physically building autonomous embodied agents, and developing a new generation of robust low-power, and autonomous real-time neural computing systems.

The results will lead to compact, low-power, intelligent sensory-motor systems that will have a large impact on service and customer robotics, the Internet of Things, and on prosthetics and personalized medicine.