Distributed Safe Navigation for Multi-Robot Systems With Fixed-Time Convergence and Deadlock Resolution

Abstract

This letter addresses the problem of distributed safe navigation for multi-robot systems (MRSs), proposing a comprehensive framework that unifies safety-critical control, real-time navigation, and deadlock resolution. To enable a fully distributed implementation of control barrier function (CBF)-based safety, this work employs a regularized primal decomposition to decouple inter-agent safety constraints. A novel distributed Fixed-Time Stable Gradient Flow (FxTS-GF) algorithm is then designed, which enables each robot to complete the optimal safety responsibility allocation and efficiently solve its local quadratic program (QP) for collision avoidance within each control cycle. Furthermore, the framework incorporates a distributed deadlock resolution mechanism that breaks undesired equilibria through coordination among neighboring robots. Theoretical guarantees are provided for the framework's anytime-safety, optimality, and fixed-time stability. Finally, the effectiveness and practical applicability of the proposed method are validated through both simulations and hardware experiments on a team of mobile vehicles.