| 一种基于强化学习的自适应变步长路径规划算法 |
| |
| 引用本文： | 王维,禹建丽.一种基于强化学习的自适应变步长路径规划算法[J].河南科技大学学报(自然科学版),2004,25(1):50-52. |
| |
| 作者姓名： | 王维 禹建丽 |
| |
| 作者单位： | 1. 河南科技大学,电子信息工程学院,河南,洛阳,471003
2. 河南科技大学,数理系,河南,洛阳,471003 |
| |
| 基金项目： | 河南省教育厅自然科学基金资助项目(20011200001) |
| |
| 摘 要： | 在基于神经网络结构的机器人全局路径规划算法中，利用强化学习的思想，引进评价预测学习的自适应变步长算法，实现了步长的在线自动调节，加快了路径规划的计算速度。仿真试验表明，该算法能有效实现步长参数的在线自动调节，并使路径规划收敛速度平均提高了10倍以上。
|
| 关 键 词： | 强化学习 自适应 变步长路径规划 算法 机器人 路 |
| 文章编号： | 1672-6871(2004)01-0050-03 |
| 修稿时间： | 2003年8月28日 |
An Adaptive Variable Stepsize Algorithm for Path Planning Based on Reinforcement Learning |
| |
| WANG Wei,YU Jian-Li.An Adaptive Variable Stepsize Algorithm for Path Planning Based on Reinforcement Learning[J].Journal of Henan University of Science & Technology:Natural Science,2004,25(1):50-52. |
| |
| Authors: | WANG Wei YU Jian-Li |
| |
| Institution: | WANG Wei~1,YU Jian-Li~2 |
| |
| Abstract: | Reinforcement learning is an important class of learning techniques that learn to perform a certain task through trial and error interactions with an knowledge-poor environment.The problem of adaptive variable stepsize of robotic path planning is studied. The method proposed here using reinforcement learning can perform on-line adaptive variance of stepsizes, and the convergence speed of path planning is increased greatly. Much simulation experiments show that the algorithm proposed can not only perform on-line adaptive variance of stepsizes effectively but also improve the convergence speed of path planning ten times than before. |
| |
| Keywords: | Exaggerated tests Global fields Paths Adaptive Control Robots |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
