Online Hierarchical Policy Learning using Physics Priors for Robot Navigation in Unknown Environments

2025-10-01

AI Summary

Robot navigation in unknown, complex indoor environments remains challenging due to the need for simultaneous exploration, mapping, and real-time path planning under physical constraints. Method: This paper proposes an online hierarchical policy learning framework integrating physics-informed priors. The high-level module constructs a sparse topological graph to model global connectivity, while the low-level module employs neural fields to solve the Eikonal partial differential equation for local obstacle avoidance and time-optimal path generation. Crucially, the framework couples neural representations with physical constraints—including time optimality and motion continuity—and leverages hierarchical decomposition to mitigate spectral bias and catastrophic forgetting. Contribution/Results: The method achieves end-to-end exploration, mapping, and navigation optimization using only online local observations—no expert demonstrations required. Evaluated on large-scale real-world and simulated environments, it significantly outperforms state-of-the-art sampling-based, imitation-learning, and ANTFields approaches, demonstrating superior accuracy, adaptability to dynamic environments, and real-time deployability.

Technology Category

• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search
• Intelligent Robots: Motion and Path Planning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments

Abstract

Robot navigation in large, complex, and unknown indoor environments is a challenging problem. The existing approaches, such as traditional sampling-based methods, struggle with resolution control and scalability, while imitation learning-based methods require a large amount of demonstration data. Active Neural Time Fields (ANTFields) have recently emerged as a promising solution by using local observations to learn cost-to-go functions without relying on demonstrations. Despite their potential, these methods are hampered by challenges such as spectral bias and catastrophic forgetting, which diminish their effectiveness in complex scenarios. To address these issues, our approach decomposes the planning problem into a hierarchical structure. At the high level, a sparse graph captures the environment's global connectivity, while at the low level, a planner based on neural fields navigates local obstacles by solving the Eikonal PDE. This physics-informed strategy overcomes common pitfalls like spectral bias and neural field fitting difficulties, resulting in a smooth and precise representation of the cost landscape. We validate our framework in large-scale environments, demonstrating its enhanced adaptability and precision compared to previous methods, and highlighting its potential for online exploration, mapping, and real-world navigation.

Problem

• Overcoming spectral bias and catastrophic forgetting in robot navigation
• Decomposing planning into hierarchical structure with physics priors
• Enabling online exploration in large unknown environments without demonstrations

Innovation

• Hierarchical planning with sparse global graphs
• Neural field planner solving Eikonal PDE locally
• Physics-informed strategy overcoming spectral bias