Cascaded Diffusion Models for Neural Motion Planning

2025-05-21

AI Summary

To address the challenge of collision constraint violations in global motion planning for robots operating in complex, cluttered environments, this paper proposes a cascaded hierarchical diffusion model. The first stage generates a topologically feasible coarse global trajectory, while the second stage refines it locally using multi-scale features. An online collision detection and re-optimization mechanism is further integrated to strictly enforce global geometric constraints. This work represents the first application of diffusion-based policies to end-to-end collision-free global trajectory generation. Evaluated on navigation and dexterous manipulation tasks, the method achieves approximately 5% higher success rates compared to state-of-the-art baselines. It significantly improves trajectory safety, feasibility, and generalization across diverse environments and task configurations.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Intelligent Robots: Motion and Path Planning
• Computer Vision: Diffusion Models for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Robots in the real world need to perceive and move to goals in complex environments without collisions. Avoiding collisions is especially difficult when relying on sensor perception and when goals are among clutter. Diffusion policies and other generative models have shown strong performance in solving local planning problems, but often struggle at avoiding all of the subtle constraint violations that characterize truly challenging global motion planning problems. In this work, we propose an approach for learning global motion planning using diffusion policies, allowing the robot to generate full trajectories through complex scenes and reasoning about multiple obstacles along the path. Our approach uses cascaded hierarchical models which unify global prediction and local refinement together with online plan repair to ensure the trajectories are collision free. Our method outperforms (by ~5%) a wide variety of baselines on challenging tasks in multiple domains including navigation and manipulation.

Problem

• Robots need collision-free motion in complex environments
• Global motion planning with multiple obstacle reasoning
• Overcoming subtle constraint violations in challenging tasks

Innovation

• Cascaded hierarchical models for motion planning
• Combines global prediction with local refinement
• Online plan repair ensures collision-free trajectories