SADP: Subgoal-Aware Diffusion Policy for Explainable Robots Learned from Foundation Model Generated Demonstrations

TL;DR

This paper introduces SADP, a diffusion policy framework that uses foundation models to generate subgoal-annotated demonstrations, enabling robots to explain their decision process and improve task success.

Contribution

SADP is the first framework to incorporate subgoal-aware diffusion policies trained on foundation model-generated demonstrations for explainable robot manipulation.

Findings

SADP outperforms task-conditioned baselines in success rates.

Provides subgoal-level signals for progress monitoring.

Achieves high task success with built-in interpretability.

Abstract

Explainable robots require not only successful task execution but also the ability to expose internal decision-making process in a user-friendly manner. However, most imitation learning methods are trained solely on task-level demonstrations, without explicitly modeling subgoal structure or execution progress. This limitation is further exacerbated by the scarcity of subgoal-level supervision in standard robot learning datasets, which restricts the development of robots that can convey the subtasks they are executing during long-horizon manipulation. To address this issue, this paper proposes Subgoal-Aware Diffusion Policy (SADP), a framework that leverages foundation models to autonomously generate subgoal-annotated demonstrations and trains diffusion policies on these datasets. SADP structures policy execution around human-interpretable subgoals by conditioning action generation on both task-level and subgoal-level descriptions. A lightweight auxiliary head further predicts subgoal completion states, allowing the robot to expose its current execution stage and monitor subgoal progression. Experiments in RLBench simulations and real-world evaluations on a UR5e robot demonstrate that SADP achieves higher task success rates than strong task-conditioned diffusion baselines, while providing subgoal-level execution signals for monitoring progress and diagnosing failures. These results highlight that built-in, rather than post-hoc, interpretability can coexist with high task performance.