Weakly-Supervised Learning for Physics-Informed Neural Motion Planning Via Sparse Roadmap
Ruiqi Ni, Yuchen Liu, Ahmed H. Qureshi
AI summary
Problem
Physics-informed neural planners struggle to scale in cluttered, multi-room environments because PDE regularization alone fails to propagate information over long horizons, often collapsing into local minima.
Approach
H-NTFields integrates weak topological bounds from a sparse roadmap with physics-informed PDE losses to anchor global structure while enforcing local geometric fidelity during training.
Key results
- Substantially higher success rates than PDE-only or roadmap-only baselines
- Reliable scaling to multi-room Gibson environments and high-DOF manipulators
- Fast amortized inference via continuous value representation integrated with sampling-based MPC
- Stabilized training through a causality-based curriculum aligning weak bounds with PDE losses
Why it matters
Enables efficient, scalable neural motion planning for complex real-world robotics without relying on expensive expert demonstrations.
Abstract
The motion planning problem requires find- ing a collision-free path between start and goal configura- tions in high-dimensional, cluttered spaces. Recent learning- based methods offer promising solutions, with self-supervised physics-informed approaches such as Neural Time Fields (NT- Fields) solving the Eikonal equation to learn value functions without expert demonstrations. However, existing physics- informed methods struggle to scale in complex, multi-room environments, where simply increasing the number of samples cannot resolve local minima or guarantee global consistency. We propose Hierarchical Neural Time Fields (H-NTFields), a weakly-supervised framework that combines weak supervision from sparse roadmaps with physics-informed PDE regular- ization. The roadmap provides global topological anchors through upper and lower bounds on travel times, while PDE losses enforce local geometric fidelity and obstacle-aware propagation. Experiments on 18 Gibson environments and real robotic platforms show that H-NTFields substantially improves robustness over prior physics-informed methods, while enabling fast amortized inference through a continuous value representation.