Train-Once Plan-Anywhere Kinodynamic Motion Planning via Diffusion Trees

Yaniv Hassidof, Tom Jurgenson, Kiril Solovey
Proceedings of The 9th Conference on Robot Learning, PMLR 305:1847-1878, 2025.

Abstract

Kinodynamic motion planning is concerned with computing collision-free trajectories while abiding by the robot’s dynamic constraints. This critical problem is often tackled using sampling-based planners (SBPs) that explore the robot’s high-dimensional state space by constructing a search tree via action propagations. Although SBPs can offer global guarantees on completeness and solution quality, their performance is often hindered by slow exploration due to uninformed action sampling. Learning-based approaches can yield significantly faster runtimes, yet they fail to generalize to out-of-distribution (OOD) scenarios and lack critical guarantees, e.g., safety, thus limiting their deployment on physical robots. We present Diffusion Tree (DiTree): a provably-generalizable framework leveraging diffusion policies (DPs) as informed samplers to efficiently guide state-space search within SBPs. DiTree combines DP’s ability to model complex distributions of expert trajectories, conditioned on local observations, with the completeness of SBPs, to yield provably-safe solutions within a few action propagation iterations for complex dynamical systems. We demonstrate DiTree’s power with an implementation combining the popular RRT planner with a DP action sampler trained on a single environment. In comprehensive evaluations on OOD scenarios, DiTree has comparable runtimes to a standalone DP (4x faster than classical SBPs), while improving the success rate over DP and SBPs (on average).

Cite this Paper

BibTeX
@InProceedings{pmlr-v305-hassidof25a, title = {Train-Once Plan-Anywhere Kinodynamic Motion Planning via Diffusion Trees}, author = {Hassidof, Yaniv and Jurgenson, Tom and Solovey, Kiril}, booktitle = {Proceedings of The 9th Conference on Robot Learning}, pages = {1847--1878}, year = {2025}, editor = {Lim, Joseph and Song, Shuran and Park, Hae-Won}, volume = {305}, series = {Proceedings of Machine Learning Research}, month = {27--30 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v305/main/assets/hassidof25a/hassidof25a.pdf}, url = {https://proceedings.mlr.press/v305/hassidof25a.html}, abstract = {Kinodynamic motion planning is concerned with computing collision-free trajectories while abiding by the robot’s dynamic constraints. This critical problem is often tackled using sampling-based planners (SBPs) that explore the robot’s high-dimensional state space by constructing a search tree via action propagations. Although SBPs can offer global guarantees on completeness and solution quality, their performance is often hindered by slow exploration due to uninformed action sampling. Learning-based approaches can yield significantly faster runtimes, yet they fail to generalize to out-of-distribution (OOD) scenarios and lack critical guarantees, e.g., safety, thus limiting their deployment on physical robots. We present Diffusion Tree (DiTree): a provably-generalizable framework leveraging diffusion policies (DPs) as informed samplers to efficiently guide state-space search within SBPs. DiTree combines DP’s ability to model complex distributions of expert trajectories, conditioned on local observations, with the completeness of SBPs, to yield provably-safe solutions within a few action propagation iterations for complex dynamical systems. We demonstrate DiTree’s power with an implementation combining the popular RRT planner with a DP action sampler trained on a single environment. In comprehensive evaluations on OOD scenarios, DiTree has comparable runtimes to a standalone DP (4x faster than classical SBPs), while improving the success rate over DP and SBPs (on average).} }
Endnote
%0 Conference Paper %T Train-Once Plan-Anywhere Kinodynamic Motion Planning via Diffusion Trees %A Yaniv Hassidof %A Tom Jurgenson %A Kiril Solovey %B Proceedings of The 9th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Joseph Lim %E Shuran Song %E Hae-Won Park %F pmlr-v305-hassidof25a %I PMLR %P 1847--1878 %U https://proceedings.mlr.press/v305/hassidof25a.html %V 305 %X Kinodynamic motion planning is concerned with computing collision-free trajectories while abiding by the robot’s dynamic constraints. This critical problem is often tackled using sampling-based planners (SBPs) that explore the robot’s high-dimensional state space by constructing a search tree via action propagations. Although SBPs can offer global guarantees on completeness and solution quality, their performance is often hindered by slow exploration due to uninformed action sampling. Learning-based approaches can yield significantly faster runtimes, yet they fail to generalize to out-of-distribution (OOD) scenarios and lack critical guarantees, e.g., safety, thus limiting their deployment on physical robots. We present Diffusion Tree (DiTree): a provably-generalizable framework leveraging diffusion policies (DPs) as informed samplers to efficiently guide state-space search within SBPs. DiTree combines DP’s ability to model complex distributions of expert trajectories, conditioned on local observations, with the completeness of SBPs, to yield provably-safe solutions within a few action propagation iterations for complex dynamical systems. We demonstrate DiTree’s power with an implementation combining the popular RRT planner with a DP action sampler trained on a single environment. In comprehensive evaluations on OOD scenarios, DiTree has comparable runtimes to a standalone DP (4x faster than classical SBPs), while improving the success rate over DP and SBPs (on average).
APA
Hassidof, Y., Jurgenson, T. & Solovey, K.. (2025). Train-Once Plan-Anywhere Kinodynamic Motion Planning via Diffusion Trees. Proceedings of The 9th Conference on Robot Learning, in Proceedings of Machine Learning Research 305:1847-1878 Available from https://proceedings.mlr.press/v305/hassidof25a.html.

Related Material