Continual Reinforcement Learning deployed in Real-life using Policy Distillation and Sim2Real Transfer

René Traoré*, Hugo Caselles-Dupré*, Timothée Lesort*, Te Sun, Natalia Diaz-Rodriguez, David Filliat

Keywords: Continual Learning, Reinforcement Learning, Sim2Real, Robotics, State Representation Learning

Abstract: We focus on the problem of teaching a robot to solve tasks presented sequentially, i.e., in a continual learning scenario. The robot should be able to solve all tasks it has encountered, without forgetting past tasks. We provide preliminary work on applying Reinforcement Learning to such setting, on navigation tasks for a 3 wheel omni-directional robot. Our approach takes advantage of state representation learning and policy distillation. Policies are trained using learned features as input, rather than raw observations, allowing for better sample efficiency. Policy distillation is used to combine different policies into a single policy that solves all encountered tasks.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/continual-reinforcement-learning-deployed-in/code)