Hybrid Coordination Framework for Centralized Task Allocation and Execution in Heterogeneous Multi-Robot Systems

Abstract

Multi-robot coordination has emerged as a critical enabler for efficient warehouse logistics, yet existing approaches struggle to balance solution quality, computational efficiency, and real-time adaptability in heterogeneous robot environments. This paper presents a novel centralized coordination architecture that integrates a hybrid optimization algorithm combining the shapley value clustering algorithm (SVCA) with the non-dominated sorting genetic algorithm II (NSGA-II) for multi-robot task allocation (MRTA). The proposed system addresses the complex challenge of allocating logistics tasks to heterogeneous mobile robots with varying payload capacities, energy consumption rates, and task deadline while simultaneously optimizing multiple conflicting objectives including makespan, energy consumption, deadline adherence, and workload balance. Unlike traditional static scheduling methods, the proposed system includes dynamic task reallocation capabilities that enable robots to autonomously detect and execute pending tasks nearby upon completion, be it with intelligent battery management that triggers autonomous recharging when energy thresholds are reached. Implemented using robot operating system 2 (ROS2) Humble and validated in Gazebo simulation environments with warehouse scenarios involve five heterogeneous robots and multiple logistics tasks. Experimental results show that, with the hybrid algorithm, 100% task allocation can be achieved with an average allocation time of 0.122 seconds, significantly outperforming state-of-the-art methods including island model genetic algorithm (IMGA), SVCA, particle swarm optimization (PSO), genetic algorithm (GA), and standalone NSGA-II executions. The system effectively manages complex logistics operations to automate the warehouse within the simulated environment, while maintaining required constraint.