运动规划
计算机科学
路径(计算)
适应(眼睛)
蒙特卡罗方法
蒙特卡罗树搜索
人工神经网络
决策树
树(集合论)
路径长度
数学优化
算法
人工智能
机器人
数学
计算机网络
数学分析
统计
物理
光学
作者
Lu Liu,Chen Wang,Fei Teng,Tieshan Li
标识
DOI:10.1109/icist59754.2023.10367178
摘要
To solve the path planning problem of finding the optimal path for a ship in a complex navigation environment, this paper uses the AlphaZero algorithm. A sufficient number of paths can be selected from the replay buffer to pursue a higher cumulative reward value and obtain the best decision policy to improve the security and efficiency of navigation through neural network training and Monte Carlo Tree Search. By observing the experimental simulation results, it is found that the AlphaZero algorithm is more adaptable and accurate in policy evaluation, which improves the security and efficiency of navigation. AlphaZero is equipped with more adaptation, policy evaluation and the capacity of path planning is improved to a higher degree.
科研通智能强力驱动
Strongly Powered by AbleSci AI