NeHMO: Neural Hamilton-Jacobi Reachability Learning for Decentralized Safe Multi-Arm Motion Planning

2026-06-30

AI Summary

This work addresses the challenges of safety and scalability in decentralized motion planning for multi-arm systems, which arise from high-dimensional coupled configuration spaces, complex collision constraints, and the unpredictable behavior of other agents. To overcome these issues, the paper proposes a safety-aware planning approach based on neural Hamilton-Jacobi reachability (HJR) analysis. It introduces neural HJR to multi-arm settings for the first time, leveraging neural networks to approximate the worst-case safety value function and embedding this within a decentralized trajectory optimization framework. This formulation enables robust, real-time safe planning without requiring explicit coordination protocols or predictions of other agents’ actions. Experimental results demonstrate that the method significantly outperforms state-of-the-art approaches in complex tasks, exhibiting strong scalability, cross-system generalization, and computational efficiency.

Abstract

Safe multi-arm motion planning is a challenging problem in robotics due to its high dimensionality, coupled configuration space, and complex collision constraints. Centralized planners are capable of coordinating all arms but often face scalability limitations, restricting applicability in real-time settings. On the other hand, decentralized methods are scalable and recent deep learning-based approaches have shown promising results. However, these depend on accurate behavior prediction or coordination protocols and may fail when other arms act unpredictably. To address these challenges, we introduce a neural Hamilton-Jacobi Reachability (HJR) learning-based approach to approximate a safety value function that captures worst-case inter-arm safety constraints. We further develop a decentralized trajectory optimization framework that uses the learned HJR representation for real-time planning. The proposed method is scalable and data-efficient, generalizes across multi-manipulator systems, and outperforms state-of-the-art baselines on challenging multi-arm motion planning tasks.

Problem

• multi-arm motion planning
• decentralized planning
• safety constraints
• collision avoidance
• real-time planning

Innovation

• Neural Hamilton-Jacobi Reachability
• Decentralized Motion Planning
• Safety Value Function
• Multi-Arm Coordination
• Data-Efficient Learning