analiza aiqbench
This paper introduces SADP, a framework that integrates foundation models into diffusion policies to autonomously generate subgoal-annotated demonstrations for long-horizon robotic manipulation. By explicitly conditioning action generation on natural-language subgoals and monitoring progress via auxiliary heads, the method achieves competitive performance while providing transparent stage-level execution signals without degrading task accuracy. The approach addresses a critical bottleneck in imitation learning where standard datasets lack explicit supervision of intermediate objectives necessary for complex coordination.
SADP: Subgoal-Aware Diffusion Policy for Long-Horizon Manipulation Learned from Foundation Model Generated Demonstrations
arXiv:2605.16871v2 Announce Type: replace Abstract: Long-horizon robot manipulation requires policies to coordinate multiple intermediate subgoals and determine when to advance between them. However, most imitation learning methods are trained solely on task-level demonstrations, without explicitly modeling the active subgoal or its execution progress. This limitation is further exacerbated by the scarcity of subgoal-level supervision in standard robot learning datasets, which makes explicit subgoal-conditioned control and online transition modeling difficult to learn. 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 explicit natural-language subgoals by conditioning action generation on both task-level and subgoal-level descriptions. A lightweight auxiliary head further predicts a continuation score that drives online subgoal switching and supports stage-level progress monitoring. Experiments in RLBench simulations and real-world evaluations on a UR5e robot demonstrate that SADP maintains competitive task performance while exposing temporally aligned subgoal-level execution signals for progress monitoring. These results show that explicit subgoal progression can be incorporated into a diffusion policy without degrading task-level performance.
Abstract:Long-horizon robot manipulation requires policies to coordinate multiple intermediate subgoals and determine when to advance between them. However, most imitation learning methods are trained solely on task-level demonstrations, without explicitly modeling the active subgoal or its execution progress. This limitation is further exacerbated by the scarcity of subgoal-level supervision in standard robot learning datasets, which makes explicit subgoal-conditioned control and online transition modeling difficult to learn. 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 explicit natural-language subgoals by conditioning action generation on both task-level and subgoal-level descriptions. A lightweight auxiliary head further predicts a continuation score that drives online subgoal switching and supports stage-level progress monitoring. Experiments in RLBench simulations and real-world evaluations on a UR5e robot demonstrate that SADP maintains competitive task performance while exposing temporally aligned subgoal-level execution signals for progress monitoring. These results show that explicit subgoal progression can be incorporated into a diffusion policy without degrading task-level performance.
Robotics (cs.RO)
arXiv:2605.16871 [cs.RO]
(or arXiv:2605.16871v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2605.16871 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 269 cuvinte
Subiecte conexe
SkillWeave: Weaving Heterogeneous Demonstrations into Long-Horizon Manipulation Skills
arXiv · 9 oct. 2026
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · 17 sept. 2026
Goal2Skill: Long-Horizon Manipulation with Adaptive Planning and Reflection
arXiv · 14 oct. 2026
Towards Neuro-Symbolic Procedural Reasoning for Long-Horizon Vision-Language-Action Manipulation
arXiv · 7 sept. 2026