神经网络在移动机器人路径规划中的应用

The Application of Neural Network in Mobile Robot Path Planning

对于环境信息完全已知的移动机器人的全局路径规划问题,应用了一种并联的神经网络结构与模拟退火算法相结合的方法,并提出了一种局部路径修正算法,最终得到一条最优的平滑路径。计算机仿真研究表明,该算法计算简单,收敛速度快,规划的路径为一条最短无碰且安全的平滑路径。在计算机仿真验证的基础上,以P3-AT型轮式移动机器人为平台,通过机器人模拟实验验证了该算法的有效性。

In this paper, for the problem of mobile robot globle path planning in an environment filled with obstacles whose shapes and positions were known, a kind of paratactically-connected neural networks combined with the Simulated Annealing was used, and brought forward a means of modifying the local path. At last, the globe optimal smooth course was achieved. The computer simulation results showed that the computation is simple, the convergence is fast and the constructed path is optimal, safe and smooth. In addition,simulated experiment on P3-AT wheeled mobile robot indicated that methods above on can get satisfying results.