论文部分内容阅读
机器人为实现在未知环境下的探索任务,必须具有自主学习其行为策略的能力.本文提出了一种自主机器人行为学习机制.机器人通过与环境的交互,基于Q-学习进行行为的自主学习.为降低学习时的计算复杂度,状态空间通过分段映射为不同的类别,从而减少状态—动作对的数量.自主机器人在未知环境中的行为学习是增量式的过程,本文将基于案例的学习与Q-学习结合,使机器人在试错时获得的经验以案例的形式保存,并实现案例库的动态更新.相关案例同时可以降低机器人行为学习时的计算复杂度和试错时的风险.在文中的最后给出了仿真结果.
In order to realize the exploration task under the unknown environment, the robot must have the ability to learn its behavior strategy autonomously.In this paper, a autonomous robot behavior learning mechanism is proposed.The robot learns the behavior based on the Q-learning through the interaction with the environment. Reduce the computational complexity of learning, the state space is mapped to different categories by segment, so as to reduce the number of state-action pairs.The behavior learning of autonomous robots in unknown environment is an incremental process, this article will be based on case study In combination with Q-learning, the experience gained by the robot in trial and error preservation is saved in the form of case and the dynamic update of the case base is also achieved.The related case can reduce the computational complexity and risk of trial-and-error during robot behavior learning. Finally, the simulation results are given.