摘要
机器人的路径规划一直是机器人研究领域的难点问题。针对煤矿井下环境的不确定性,环境的复杂使机器人很难得到好的规划结果。采用强化学习算法中的Q-learning算法实现井下移动机器人的局部路径规划,并对Q函数中的即时回报进行加权修正,使算法更有效地利用环境特征信息,进一步提高了避障能力。最后通过VC++进行仿真和模拟。仿真实验说明该方法的有效性和可行性。
Path planning of robot is still a difficult question in the robot research domain. Q- learning algorithm is used to realize local path planning of mobile robot under coal mine for environment uncertainty of coal mine, because it is difficult to obtain a good path in such a complex enivronment. Instant rewards in the function are modified by using weight,then the information of environment characteristics is used effectively to avoid the obstacles. At last the algorithm is simulated by using VC＋＋. Simulation shows that this algorithm is efficient and feasible.
出处
《现代电子技术》
2008年第24期106-108,共3页
Modern Electronics Technique
关键词
移动机器人
不确定环境
强化学习
路径规划
mobile robot
uncertain environment
reinforcement learning
path planning
