GNN-DIP: Neural Corridor Selection for Decomposition-Based Motion Planning

2026-03-12

AI Summary

This work addresses the inefficiency of sampling-based motion planners in narrow passages and the combinatorial explosion of candidate corridors that hinders hierarchical decomposition approaches. To overcome these limitations, we propose a novel method integrating graph neural networks (GNNs) with a two-stage hierarchical planner. The GNN predicts doorway scores on a cell adjacency graph to guide an efficient and complete corridor search. Our pipeline further combines constrained Delaunay triangulation, Slab-based convex decomposition, and the Funnel algorithm to generate near-optimal paths in both 2D and 3D environments. This is the first approach to employ GNNs for guiding corridor selection in hierarchical planning, achieving significant speedups while preserving completeness. Experiments demonstrate 99–100% success rates in challenging scenarios—including 2D narrow passages, a 3D bottleneck with 246 obstacles, and dynamic 2D environments—with runtime improvements of 2× to 280× over sampling-based baselines.

Technology Category

• Intelligent Robots: Motion and Path Planning
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Motion planning through narrow passages remains a core challenge: sampling-based planners rarely place samples inside these narrow but critical regions, and even when samples land inside a passage, the straight-line connections between them run close to obstacle boundaries and are frequently rejected by collision checking. Decomposition-based planners resolve both issues by partitioning free space into convex cells -- every passage is captured exactly as a cell boundary, and any path within a cell is collision-free by construction. However, the number of candidate corridors through the cell graph grows combinatorially with environment complexity, creating a bottleneck in corridor selection. We present GNN-DIP, a framework that addresses this by integrating a Graph Neural Network (GNN) with a two-phase Decomposition-Informed Planner (DIP). The GNN predicts portal scores on the cell adjacency graph to bias corridor search toward near-optimal regions while preserving completeness. In 2D, Constrained Delaunay Triangulation (CDT) with the Funnel algorithm yields exact shortest paths within corridors; in 3D, Slab convex decomposition with portal-face sampling provides near-optimal path evaluation. Benchmarks on 2D narrow-passage scenarios, 3D bottleneck environments with up to 246 obstacles, and dynamic 2D settings show that GNN-DIP achieves 99--100% success rates with 2--280 times speedup over sampling-based baselines.

Problem

• motion planning
• narrow passages
• decomposition-based planning
• corridor selection
• combinatorial explosion

Innovation

• Graph Neural Network
• Decomposition-based Planning
• Narrow Passage
• Corridor Selection
• Motion Planning