NBDI: A Simple and Effective Termination Condition for Skill Extraction from Task-Agnostic Demonstrations

TL;DR

This paper introduces NBDI, a novel method for identifying decision points in skill learning using a state-action novelty module, which improves performance in complex, long-horizon tasks and adapts well to environment variations.

Contribution

The paper proposes NBDI, a simple termination condition based on state-action novelty, enhancing skill extraction and decision point identification in long-horizon, variable environments.

Findings

NBDI outperforms previous methods in complex tasks.

NBDI remains effective despite environment variations.

Improves exploration and policy learning efficiency.

Abstract

Intelligent agents are able to make decisions based on different levels of granularity and duration. Recent advances in skill learning enabled the agent to solve complex, long-horizon tasks by effectively guiding the agent in choosing appropriate skills. However, the practice of using fixed-length skills can easily result in skipping valuable decision points, which ultimately limits the potential for further exploration and faster policy learning. In this work, we propose to learn a simple and effective termination condition that identifies decision points through a state-action novelty module that leverages agent experience data. Our approach, Novelty-based Decision Point Identification (NBDI), outperforms previous baselines in complex, long-horizon tasks, and remains effective even in the presence of significant variations in the environment configurations of downstream tasks, highlighting the importance of decision point identification in skill learning.