Research on escape strategy of unmanned surface vehicle in pursuit evasion problem
-
摘要: 摘 要：[ 目的 ] 针对敌方船舶采用合围战术，本文研究了我方无人艇（Unmanned Surface Vehicle，USV）被敌方船舶包围的情况下的逃跑策略规划问题。[ 方法 ] 本文提出了一种混合采样深度Q网络（Hybrid Sampling Deep Q Network，HS-DQN）强化学习算法，设计了状态空间，动作空间和奖励函数，通过训练获得最优的USV逃跑策略，并从奖励值和逃跑成功率方面与DQN算法进行对比。[ 结果 ] 仿真结果表明，使用HS-DQN算法进行训练，逃跑成功率提高2.00%，算法的收敛速度提高了20%。[ 结论 ] HS-DQN算法可以减少USV无效探索的次数，并加快算法的收敛速度，仿真试验验证了USV逃跑策略的有效性。Abstract: abstract: [Objectives] Aiming at the encirclement tactics adopted by enemy ships, the problem of planning the escape strategy are studied when unmanned surface vehicle is surrounded by enemy ships, in this paper. [Methods] A hybrid sampling deep Q-network (HS-DQN) reinforcement learning algorithm is proposed and the state space, action space and reward function are designed to obtain the optimal escape strategy of USV. It is compared with the DQN algorithm in terms of reward and escape success rate. [Results] The simulation results show that using the HS-DQN algorithm for training, the escape success rate is increased by 2.00%, and the convergence speed of the algorithm is increased by 20%. [Conclusion] The HS-DQN algorithm can reduce the number of USV useless explorations and speed up the convergence of the algorithm. And the simulation experiments verify the effectiveness of the escape strategy of USV.
-
计量
- 文章访问数: 261
- 被引次数: 0
下载:
