SPACE-TIME MULTI-AGENT ROUTE PLANNING IN CONDITIONS OF LIMITED RESOURCES

Abstract

In modern industrial and logistics environments, efficient operation of Autonomous Guided Vehicles (AGVs) and Mobile Robotic Platforms (MRPs) critically depends on their ability to plan and adapt trajectories in dynamic, multi-agent settings while respecting energy constraints. This paper presents a federated multi-agent spatio-temporal path planning algorithm that integrates local autonomous planning with iterative coordination among agents to avoid collisions and deadlocks. The proposed method incorporates dynamic obstacles, static constraints, and residual battery levels directly into the cost function, enabling energy-aware route selection. Simulation results for four agents demonstrate that all agents successfully reach their goals while maintaining battery levels above minimum thresholds. The algorithm achieves effective collision avoidance, minimizes waiting times, and balances path length with energy efficiency. Compared to traditional centralized methods, the federated approach reduces computational complexity and scales efficiently with the number of agents. These results highlight the practical applicability of the method for real-time, energy-constrained multi-robot operations in industrial settings.