| 动态环境中基于增强式学习的路径规划方法 |
| |
| 引用本文： | 庄晓东,孟庆春,熊建设,殷波,王汉萍.动态环境中基于增强式学习的路径规划方法[J].机器人,2001(Z1). |
| |
| 作者姓名： | 庄晓东 孟庆春 熊建设 殷波 王汉萍 |
| |
| 作者单位： | 青岛海洋大学电子工程系 青岛海洋大学电子工程系 青岛 |
| |
| 基金项目： | 高等学校重点实验室访问学者基金资助 |
| |
| 摘 要： | 本文结合机器人路径规划问题介绍了增强式学习方法 ,实现了动态环境中基于增强式学习的自适应路径规划 .增强式学习通过采用随机性的控制策略 ,实现策略的优化搜索和在线学习 .并采用具有模式增强输入的BP网络进行决策参数估计 ,加快学习的收敛 .仿真试验证明该方法能有效实现动态环境中机器人的避碰和导航
|
| 关 键 词： | 动态环境 路径规划 增强式学习 自适应 |
A METHOD OF PATH PLANNING IN DYNAMIC ENVIRONMENT BASED ON REINFORCEMENT LEARNING |
| |
| ZHUANG Xiao,dong\,MENG Qing,chun\,XIONG Jian,she\ YIN Bo\ WANG Han,ping.A METHOD OF PATH PLANNING IN DYNAMIC ENVIRONMENT BASED ON REINFORCEMENT LEARNING[J].Robot,2001(Z1). |
| |
| Authors: | ZHUANG Xiao dong\ MENG Qing chun\ XIONG Jian she\ YIN Bo\ WANG Han ping |
| |
| Abstract: | To solve the robot's path planning problem in dynamic environment, this paper applies adaptive learning to path planning based on reinforcement learning. A strategy with stochastic property is used in reinforcement learning to perform an optimal search in the strategy space. BP neural networks with pattern extended input are used to estimate control parameters, and the learning speed is increased. Computer simulation experiment proves the method's validity for robot's obstacle avoidance and navigation in dynamic environment. |
| |
| Keywords: | dynamic environment path planning reinforcement learning adaptive learning |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
