Memory Anchors for Continual Robot Learning

Maximilian Du, Zhanyi Sun, Chen Xu +3

Robot policies deployed in the wild should have the capability to continually learn new tasks without forgetting existing behaviors. A common approach to combat such catastrophic f…