摘要：

因RoboCup四腿组比赛所采用机器人的关节刚度较低，采用固定形状行走轨迹的步态规划方法所产生的实际步态和规划步态偏差较大。这种偏差限制了机器人行走速度的提高。为了提高机器人的行走速度，在原有步态规划方法的基础上，引入200组坐标描述任意形状的机器人行走轨迹，并用改进的遗传算法寻找最适合机器人行走的轨迹形状。实验结果表明：改进的行走轨迹规划方法经学习后的实际步态更有利于机器人的行走，在行走更加平稳的同时可使机器人获得更快速的行走效果。

关键词: 步态规划, 四足机器人, RoboCup, 遗传算法

Abstract: The actual gait of robot significantly bias from the planned one for the low rigidity of the robot’s joints, which is generated by fixed shapped walk loci in the RoboCup 4-legged league, and that limits the robot walking speed. According to this, 200 set of coordinates are employed to represent the arbitrary shaped robot walk loci based on previous gait planning, and an improved GA (genetic algorithm) is employed to achieve the optimum shaped walk loci for robot walking. The experiment results show that the actual gait is more efficient for robot walking by the improved walk loci optimization and provides the robot with a fast and smooth gait.

Key words: gait planning, quadruped robot, RoboCup, genetic algorithm

中图分类号: 

• TP24