基于动作决策的机器鱼顶球算法

Behavior-Based Motion Planning of Biomimetic Robot-Fish

为了使机器鱼的顶球兼备速度性和连贯性并提高进球效率,提出一种基于动作决策的顶球算法。在分析典型的顶球算法的基础上,设计了机器鱼5种简单动作,依据机器鱼、球和对方球门的坐标几何位置关系的设计决策机制来选择机器鱼下一步动作,并在机器人水球比赛平台上进行了单鱼顶球实验进行算法验证。实验结果表明,该算法能大大缩短一次进球所需要的时间。

In order to make the ball-pushing movement fast and coherent, and to improve the efficiency of goals, on the basis of analyzing the basic ball-pushing algorithms, a action-based algorithm was proposed, which designed five simple actions and used arbiter for action decision according to the geometry sites of robotic fish, ball and opponent＇s gate. Validate the algorithm by a ball-pushing experiment on the robotic fish water polo game platform. Results show this ball-pushing algorithm can largely shorten one goal time.