Learning to Discover Skills through Guidance

TL;DR

DISCO-DANCE is a novel unsupervised skill discovery algorithm that improves exploration by guiding skills towards unexplored states, outperforming existing methods in complex environments.

Contribution

The paper introduces DISCO-DANCE, a new USD algorithm that selects guide skills and guides others to enhance exploration in complex environments.

Findings

DISCO-DANCE outperforms baseline methods in navigation and control benchmarks.

It effectively explores unexplored states in challenging environments.

Qualitative visualizations demonstrate its superior exploration capabilities.

Abstract

In the field of unsupervised skill discovery (USD), a major challenge is limited exploration, primarily due to substantial penalties when skills deviate from their initial trajectories. To enhance exploration, recent methodologies employ auxiliary rewards to maximize the epistemic uncertainty or entropy of states. However, we have identified that the effectiveness of these rewards declines as the environmental complexity rises. Therefore, we present a novel USD algorithm, skill discovery with guidance (DISCO-DANCE), which (1) selects the guide skill that possesses the highest potential to reach unexplored states, (2) guides other skills to follow guide skill, then (3) the guided skills are dispersed to maximize their discriminability in unexplored states. Empirical evaluation demonstrates that DISCO-DANCE outperforms other USD baselines in challenging environments, including two navigation benchmarks and a continuous control benchmark. Qualitative visualizations and code of DISCO-DANCE are available at https://mynsng.github.io/discodance.