Enabling Long-term Robot Autonomy through Adaptable Fault Resilience

An active Space Technology Research Grants project developing long-duration robot autonomy that adapts in real time to unforeseen system failures and wear. The work combines controller-based fault detection, isolation and recovery with redundant actuation, reinforcement learning, and task-completion probability analysis so robots can reprioritize tasks and request assistance during long surface or orbital deployments.