一种快速神经网络路径规划算法

本文研究已知障碍物形状和位置环境下的全局路径规划问题，给出了一个路径 规划算法，其能量函数利用神经网络结构定义，根据路径点位于障碍物内外的不同位置选取 不同的动态运动方程，并针对障碍物的形状设定各条边的模拟退火初始温度．仿真研究表明 ，本文提出的算法计算简单，收敛速度快，能够避免某些局部极值情况，规划的无碰路径达 到了最短无碰路径．

FAST ALGORITHM FOR PATH PLANNING BASED ON NEURAL NETWORK

In this paper, the problem of global path planning is studied for a moving robot in an environment filled with obstacles whose shapes and positions are known. An aggressive algorithm for path planning is presented. The obstacles are described by an energy function defined using neural networks; different path generating equations are used, depending on whether the path points lie inside or outside the obstacles, which allows high speed of the calculations and fast convergence. The simulation results show that the computation is simple, some local minimum problems can be avoided, and the constructed path is optimal and piecewise linear.