多移动机器人动态联盟围捕策略

Dynamic alliance pursuit strategy for multiple mobile robots

针对多移动机器人协作围捕中入侵者的位置和数量未知的问题,提出了一种基于改进合同网协议的动态联盟围捕策略。管理者通过分析入侵者的特征,与历史案例完成最佳匹配,并根据信誉度的高低,选出最为合适的承包商(即机器人小组)完成围捕任务。距离某入侵者较近的其他小组中的机器人,能够与围捕该入侵者的机器人小组形成联盟,并替换掉小组中较远的机器人,而领先完成任务的小组们可与规定时间内未完成任务的小组形成联盟,以更好地完成围捕任务,并且机器人可以随时退盟。对传统的人工势场方法进行改进,解决了"目标不可到达"和"死锁"问题。面对不同数量的入侵者分别进行了仿真实验,并与经典围捕算法对比,结果表明所提围捕策略能够缩短围捕时间,顺利完成围捕任务。

Aiming at the unknown location and quantity of intruders in cooperative pursuit of multiple mobile robots, a dynamic alliance pursuit strategy based on improved contract net protocol was proposed. Firstly, the manager analyzed characteristics of intruders to complet the best matching strategy with historical cases and selected the most suitable contractor (group) to complete the pursuit mission according to the level of credibility. Secondly, the robot closer to an intruder,would ally with the group for puisuiting the intruder and the robot groups completing taskes ahead would form an alliance with the robot group failing to complete task in the given time with freedom to withdraw from the alliance at any time for all robots. Finally, the traditional artificial potential field method was improved to solve the problem of “unreachable target” and “deadlock”. The imulation experiments were carried out with different numbers of intruders and the comparison with classic pursuit algorithm was performed. The results show that the proposed strategy can shorten the pursuit time and complete the pursuit task successfully.