论文部分内容阅读
提出一种未知环境下基于有先验知识的滚动Q学习机器人路径规划算法.该算法在对Q值初始化时加入对环境的先验知识作为搜索启发信息,以避免学习初期的盲目性,可以提高收敛速度.同时,以滚动学习的方法解决大规模环境下机器人视野域范围有限以及因Q学习的状态空间增大而产生的维数灾难等问题.仿真实验结果表明,应用该算法,机器人可在复杂的未知环境中快速地规划出一条从起点到终点的优化避障路径,效果令人满意.
This paper proposes a path planning algorithm based on prior knowledge of rolling Q learning robots in unknown environment.This algorithm adds priori knowledge of environment to search enlightenment information when initializing Q to avoid the blindness in the initial stage of learning and can improve Convergence rate.At the same time, the method of rolling learning is used to solve the problem of the limited scope of robot’s field of view in large-scale enviroment and the dimensionality disaster caused by the increase of the state space of Q learning.Experimental results show that using this algorithm, In a complex and unknown environment, an optimal obstacle avoidance path from the starting point to the end point is rapidly planned, and the result is satisfactory.