基于修正的最小均方误差分类算法的移动机器人运动规划

研究了不确定性环境下移动机器人躲避运动轨迹未知的移动障碍物的一种新方法．通过实时最小均方误差估计算法预测每个障碍物的位置及运动轨迹，并利用模式识别中最小均方误差分类器的修正模型计算出机器人的局部避障路径，再运用船舶导航中使用的操纵盘技术来确定每个导航周期中移动机器人的速．度仿真结果表明了该方法的可行性

This article presents a new collision  avoidance scheme for motion planning of a mobile robot. The goal is to conduct collision  free navigation of a three-wheel mobile robot among moving obstacles with unknown trajectories. The trajectory of each obstacle is predicted by a real  time least  mean  square  error (LMSE) estimation method. Based on the predicted posi  tions of obstacles, a local collision  free path is computed by using a modified version of the LMSE classifier in pattern recognition. And the maneuvering board technique used for nauti  cal navigation is employed to determine the speed of the mobile robot for each navigation cy  cle. Simulation results are presented to show the feasibility of the proposed approach.

中国科学院机器人学开放研究实验室基金