论文部分内容阅读
强化学习一词来自于行为心理学 ,这门学科把行为学习看成反复试验的过程 ,从而把环境状态映射成相应的动作 .在设计智能机器人过程中 ,如何来实现行为主义的思想、在与环境的交互中学习行为动作 ?文中把机器人在未知环境中为躲避障碍所采取的动作看作一种行为 ,采用强化学习方法来实现智能机器人避碰行为学习 .为了提高机器人学习速度 ,在机器人局部路径规划中的状态空间量化就显得十分重要 .本文采用自组织映射网络的方法来进行空间的量化 .由于自组织映射网络本身所具有的自组织特性 ,使得它在进行空间量化时就能够较好地解决适应性灵活性问题 ,本文在对状态空间进行自组织量化的基础方法上 ,采用强化学习 .解决了机器人避碰行为的学习问题 ,取得了满意的学习结果
The term reinforcement learning comes from behavioral psychology, which treats behavioral learning as a process of trial and error, mapping the state of the environment into action.How to implement the concept of behaviorism in the design of intelligent robots, Environment interaction behavior of learning behavior? The robot robot in an unknown environment to avoid the obstacles taken by the action as a behavior, the use of enhanced learning methods to achieve intelligent robot avoid collision behavior learning.In order to improve the robot learning speed, in the robot part It is very important to quantify the state space in path planning.This paper uses the method of self-organizing mapping network to quantify the space.As the self-organizing feature of self-organizing mapping network itself, it can be better in the space quantization In order to solve the problem of adaptive flexibility, this paper applies reinforcement learning to the basic method of self-organizing quantitative state space, solves the learning problem of robot avoidance behavior, and achieves satisfactory learning results