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

2025-07-18

AI Summary

Balancing safety and scalability remains challenging in multi-agent motion planning (MAMP), particularly under high-dimensional dynamics and real-time constraints. Method: This paper introduces the first decentralized, real-time safe planning framework for MAMP based on neural Hamilton–Jacobi (HJ) reachability learning. It approximates high-dimensional HJ reachable sets via neural networks—explicitly encoding worst-case collision constraints—without inter-agent communication or behavioral prediction. A learned safety cost function is integrated with decentralized trajectory optimization, ensuring compatibility with diverse complex dynamical systems. Contribution/Results: Evaluated in high-dimensional settings—including a 12D dual-arm collaborative manipulation task—the approach significantly outperforms existing decentralized and centralized methods. It achieves high data efficiency, strong generalization across unseen scenarios, and real-time decision-making capability, establishing a new benchmark for scalable, provably safe multi-agent planning.

Technology Category

• Multiagent Systems: Multiagent Planning
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Safe Multi-Agent Motion Planning (MAMP) is a significant challenge in robotics. Despite substantial advancements, existing methods often face a dilemma. Decentralized algorithms typically rely on predicting the behavior of other agents, sharing contracts, or maintaining communication for safety, while centralized approaches struggle with scalability and real-time decision-making. To address these challenges, we introduce Neural Hamilton-Jacobi Reachability Learning (HJR) for Decentralized Multi-Agent Motion Planning. Our method provides scalable neural HJR modeling to tackle high-dimensional configuration spaces and capture worst-case collision and safety constraints between agents. We further propose a decentralized trajectory optimization framework that incorporates the learned HJR solutions to solve MAMP tasks in real-time. We demonstrate that our method is both scalable and data-efficient, enabling the solution of MAMP problems in higher-dimensional scenarios with complex collision constraints. Our approach generalizes across various dynamical systems, including a 12-dimensional dual-arm setup, and outperforms a range of state-of-the-art techniques in successfully addressing challenging MAMP tasks. Video demonstrations are available at https://youtu.be/IZiePX0p1Mc.

Problem

• Decentralized safe multi-agent motion planning scalability challenge
• High-dimensional collision and safety constraints modeling difficulty
• Real-time decentralized trajectory optimization for complex systems

Innovation

• Neural HJR modeling for high-dimensional spaces
• Decentralized trajectory optimization with HJR
• Generalizes across various dynamical systems