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

2025-08-28

AI Summary

To address the inefficiency of sampling-based planners (SBPs) and the poor generalization and lack of safety guarantees in learning-based methods under dynamical constraints, this paper proposes DiTree—a novel framework that integrates diffusion policies as directed samplers into RRT-style search trees. This unifies learning-driven exploration efficiency with the completeness and safety guarantees of classical motion planning. DiTree models expert trajectory distributions conditioned on local observations, enabling zero-shot deployment to arbitrary unseen environments after a single training phase. Experiments demonstrate that, in out-of-distribution (OOD) scenarios, DiTree achieves an average 3× speedup and ~30% higher success rate compared to conventional SBPs, while matching the inference speed of pure learning-based approaches. Crucially, all generated trajectories strictly satisfy dynamical feasibility and collision-avoidance constraints.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

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 emph{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 emph{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 emph{single environment}. In comprehensive evaluations on OOD scenarios, % DiTree has comparable runtimes to a standalone DP (3x faster than classical SBPs), while improving the average success rate over DP and SBPs. DiTree is on average 3x faster than classical SBPs, and outperforms all other approaches by achieving roughly 30% higher success rate. Project webpage: https://sites.google.com/view/ditree.

Problem

• Improve kinodynamic motion planning with diffusion-guided sampling
• Address slow exploration in sampling-based planners via learned policies
• Ensure safety and generalization in out-of-distribution scenarios

Innovation

• Diffusion Trees guide kinodynamic planning
• Combines diffusion policies with sampling-based planners
• Provably-safe solutions with few iterations