中文摘要：

针对URWPGSim2D仿真平台,为实现机器鱼快速、准确的调整,本文将机器鱼的状态定义为＂调整＂和＂推球＂,并提出基于极限学习机的动作决策模型,利用此模型自主选择相应的动作策略。动作决策模型根据当前时刻周围的环境信息,利用极限学习机确定机器鱼的状态,自主选择当前时刻的最优击球点,并确定机器鱼速度和角速度档位的最优组合。经URWPGSim2D仿真平台验证结果表明：机器鱼可根据策略调整路径,选择合适的动作策略,以更少的时间代价完成比赛。这说明基于极限学习机的动作决策策略能充分考虑机器鱼和水球的实时信息,在不同情况下选择不同的策略,具有很强的适应能力,满足仿真机器鱼对于动作决策的要求。

英文摘要：

Aiming at URWPGSim2D simulation platform, in order to realize rapid and accurate adjustment of simulation robotic fish, this paper defined the state of robotic fish for＂adjustment＂and＂push ball＂, and action decision model based on extreme learning machine is put forward.By using this model, the corresponding action strategies are selected.In the action decision model, according to the current environ- ment information around the robotic fish, the state of the robotic fish is determined by the extreme learning machine.Then the fish can indepen- dently choose the optimal hitting point of the current time, and determine the optimal combination of velocity and angular velocity.Verified by URWPGSim2D simulation platform show that：the robotic fish can choose the appropriate action strategy to adjust its path by using the action decision model, and complete the competition with less time.This shows that action decision-making strategy based on extreme learning ma- chine can fully consider the real-time information of robotic fish and water polo, choose a different strategy in different cases, have a strong ability to adapt, meet the requirements of simulation robotic fish for the action decisions.