Policy Stitching: Learning Transferable Robot Policies

Understand

Training robots with reinforcement learning (RL) typically involves heavy interactions with the environment, and the acquired skills are often sensitive to changes in task environments and robot kinematics.

• Transfer RL aims to leverage previous knowledge to accelerate learning of new tasks or new body configurations.
• However, existing methods struggle to generalize to novel robot-task combinations and scale to realistic tasks due to complex architecture design or strong regularization that limits the capacity of the learned policy.
• We propose Policy Stitching, a novel framework that facilitates robot transfer learning for novel combinations of robots and tasks.