Lifelong Learning of Perception and Action in Autonomous Systems

This DARPA-sponsored multi-university project (Brown, UMichigan, UT Austin, and USC, led by Dr. Eaton at Penn) focuses on developing a comprehensive approach to lifelong machine learning in autonomous systems.  It addresses fundamental issues of continual learning and transfer across diverse tasks, scalable knowledge maintenance, self-directed learning, and adaptation to changing environments with guaranteed performance and safety. These methods will be applied to integrated perception and action in mobile indoor service robots, yielding autonomous systems that can rapidly learn diverse tasks in unstructured environments and quickly adapt to changes.