基于分布式拍卖机制的多移动机器人动态任务分配算法

任务分配是多移动机器人调度系统的关键问题之一,为了提高任务整体完成效率,提出了一种基于分布式拍卖机制的多移动机器人动态任务分配算法.该方法对机器人群体采用分布式控制方法,彼此共享且动态更新任务集,采用分布式的拍卖机制竞拍任务,增加了调整任务执行顺序环节,考虑任务整体完成效率,最后在Linux系统下搭建了多机器人和障碍物的仿真环境.结果表明,该算法分配效率高于线性(CLP)算法和混合整数求解(CBC)算法,且具有稳定性,相比执行效率高的深度强化学习(DQN)算法和空缺链(VC)算法,执行效率稳定,移动代价降低了55%,实现了较高执行效率和低移动代价之间的平衡,可应用于实际仿真环境,具有可行性.

Dynamic task assignment algorithm for multi-mobile robots based on distributed auction mechanism

Task allocation is one of the key issues in multi-mobile robot scheduling system. The existing methods tend to focus on improving the efficiency of task assignment without considering the task execution time, ignoring the overall task completion efficiency and the computational cost. In addition, they usually adopt a centralized control architecture, which requires the robot to be in a stable and reliable communication environment. In order to solve the above problems, a distributed auction mechanism-based dynamic task assignment algorithm for multiple mobile robots is proposed. The method adopts a distributed control method for robot groups, shares and dynamically updates task sets with each other, bids for tasks using a distributed auction mechanism, adds a link to adjust the task execution order, and considers the overall task completion efficiency. Finally, a simulation environment with multiple robots and obstacles is built under Linux system and relevant experiments are conducted. The experimental results show that the algorithm has higher allocation efficiency than the linear and mixed integer solving algorithms CLP and CBC, and has stability. Comparing with the Deep Reinforcement Learning (DQN) and Vacancy Chain (VC) algorithms, the execution efficiency remains stable while the movement cost is significantly reduced by 55％, which achieves a balance between the efficiency and cost in operation. This study, as believed, can be put to the real-world applications with promising outlook.