论文部分内容阅读
在多机器人环境中,由于每个机器人动作选择的重叠现象,让机器人之间的协作变得很差.提出了一个方法用于确定动作选择级别.在此基础上,可以很好地控制多机器人的协作行为的获取.首先,定义了用于动作选择级优先级的8个级别,这8个级别相应的映射到8个动作子空间.然后,利用局部势场法,每个机器人的动作选择优先级被计算出来,并且因此,每个机器人获得了各自需要搜索的动作子空间.在动作子空间中,每个机器人利用加强学习方法来选择一个适当的动作.最终,把该方法用于机器人足球比赛的机器人局部协作训练中.试验的效果在仿真和实际比赛中得到了证实.
In a multi-robot environment, the collaboration among robots becomes poor due to the overlap of the selection of each robot motion. A method is proposed for determining the level of motion selection, on the basis of which the multi-robot First of all, it defines eight levels for action selection level priority, which are correspondingly mapped to eight action subspaces.Then, using the local potential field method, each robot’s action selection The priorities are calculated, and therefore, each robot obtains the action subspaces to search for. In the action subspace, each robot uses the reinforcement learning method to select an appropriate action. Finally, the method is applied to the robot The robotics in football match local cooperation training.The effect of experiment is confirmed in simulation and actual competition.