Merry-Go-Round: Safe Control of Decentralized Multi-Robot Systems with Deadlock Prevention

2025-03-07

AI Summary

To address safety-critical control and deadlock issues in decentralized multi-robot navigation, this paper proposes a hybrid control framework based on dynamic “ring roads.” The method relies solely on local peer-to-peer communication and basic obstacle-avoidance controllers, achieving system-level forward progress without global planning via three key mechanisms: conditional, on-demand ring formation; unidirectional circumnavigation; and autonomous ring disengagement. Its core innovation is a novel dynamic ring-road mechanism for proactive deadlock prevention, integrating reference-path-guided circular motion planning with a lightweight state-switching strategy. Extensive simulations and real-robot experiments demonstrate that, in dense and cluttered environments, the approach significantly outperforms state-of-the-art decentralized methods in task success rate and arrival rate, while incurring minimal communication and computational overhead. The framework ensures safety, scalability, and full decentralization.

Technology Category

• Intelligent Robots: Multi-Robot Systems
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Multiagent Systems: Distributed Problem Solving

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Decentralized Web and Fediverse systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

We propose a hybrid approach for decentralized multi-robot navigation that ensures both safety and deadlock prevention. Building on a standard control formulation, we add a lightweight deadlock prevention mechanism by forming temporary"roundabouts"(circular reference paths). Each robot relies only on local, peer-to-peer communication and a controller for base collision avoidance; a roundabout is generated or joined on demand to avert deadlocks. Robots in the roundabout travel in one direction until an escape condition is met, allowing them to return to goal-oriented motion. Unlike classical decentralized methods that lack explicit deadlock resolution, our roundabout maneuver ensures system-wide forward progress while preserving safety constraints. Extensive simulations and physical robot experiments show that our method consistently outperforms or matches the success and arrival rates of other decentralized control approaches, particularly in cluttered or high-density scenarios, all with minimal centralized coordination.

Problem

• Ensures safe decentralized multi-robot navigation
• Prevents deadlocks using temporary roundabout mechanisms
• Improves success rates in cluttered, high-density environments

Innovation

• Hybrid approach for decentralized multi-robot navigation
• Lightweight deadlock prevention using temporary roundabouts
• Local peer-to-peer communication for collision avoidance