Goal-Conditioned Reinforcement Learning
with Imagined Subgoals

International Conference on Machine Learning (ICML), 2021


Goal-conditioned reinforcement learning endows an agent with a large variety of skills, but it often struggles to solve tasks that require more temporally extended reasoning. In this work, we propose to incorporate imagined subgoals into policy learning to facilitate learning of complex tasks. Imagined subgoals are predicted by a separate high-level policy, which is trained simultaneously with the policy and its critic. This high-level policy predicts intermediate states halfway to the goal using the value function as a reachability metric. We don’t require the policy to reach these subgoals explicitly. Instead, we use them to define a prior policy, and incorporate this prior into a KL-constrained policy iteration scheme to speed up and regularize learning. Imagined subgoals are used during policy learning, but not during test time, where we only apply the learned policy. We evaluate our approach on complex robotic navigation and manipulation tasks and show that it outperforms existing methods by a large margin.

Method Overview


Ant Navigation

Ant U-maze navigation
Ant U-maze
Ant U-maze navigation
Ant S-maze
Ant Pi-maze environment
Ant π-maze
Ant Omega-maze environment
Ant ω-maze

Vision-Based Robotic Manipulation

Robot Illustration
Robot Illustration