Q-Learning-Based Task-Allocation for Multi-Robot Systems: From One-Line Tasks to Bi-Line Tasks
Xiulei Han, Qi Mao, Fei Xie, Jun Chen, Yijian Liu, S. Li, Wanwan Zhu · 2025
In this paper, we investigate the problem of task allocation for multiple-robot systems within the implementation of a Q-learning-based algorithm. Under the setting of the time required to finish the corresponding job of each robot, we employ a hybrid reward mechanism combining local immediate rewards with global temporal reward allocation to update the Q-table, so as to attain the target of the minimal time for task allocation. Building upon the incorporation of such ideas, we enable multi-robot task allocation ranging from one-line tasks to bi-line tasks. For more intricate dual-line tasks, we propose a weighted bi-line allocation algorithm to solve the scenario that one of the single-line tasks is in need of accelerated completion. Finally, the results of several experiments are provided to confirm the effectiveness and efficiency of the proposed task allocation algorithm.