Walk the Random Walk: Learning to Discover and Reach Goals Without Supervision

TL;DR

This paper introduces a method for training goal-conditioned agents in continuous control tasks without external rewards or domain knowledge, using random walks to learn reachability and goal memory for autonomous skill acquisition.

Contribution

The paper presents a novel approach combining reachability networks and goal memory to enable goal discovery and reaching without supervision.

Findings

Effective in continuous control navigation tasks

Achieves diverse goal reaching without external rewards

Maintains updated goal memory during learning

Abstract

Learning a diverse set of skills by interacting with an environment without any external supervision is an important challenge. In particular, obtaining a goal-conditioned agent that can reach any given state is useful in many applications. We propose a novel method for training such a goal-conditioned agent without any external rewards or any domain knowledge. We use random walk to train a reachability network that predicts the similarity between two states. This reachability network is then used in building goal memory containing past observations that are diverse and well-balanced. Finally, we train a goal-conditioned policy network with goals sampled from the goal memory and reward it by the reachability network and the goal memory. All the components are kept updated throughout training as the agent discovers and learns new goals. We apply our method to a continuous control navigation and robotic manipulation tasks.