基于激光扫描的移动机器人行走环境直线检测

提出一种移动机器人行走环境直线检测算法;对激光传感器采集的环境信息作预处理,设计分割、提取规则将离散距离信息转化为具备明显特征的直线段序列,考虑传感器误差用最小二乘法拟合直线段;计算拟合误差作为直线分割提取的阈值自动调整条件,实现阈值自动调整.基于极坐标内直线协方差矩阵计算Mahalanobis距离,实现直线合并,不丢失环境信息同时降低其直线存储量.与传统的Split-and-Merge算法相比,解决了对分割阈值参数敏感的问题.经过实验证明:直线检测算法能够有效检测出直线特征.

Straight line detection of walking environment of mobile robot based on laser scanning

A walking environment line detection algorithm for mobile robot is proposed;preprocessing on environment information is performed,segmentation and extraction regulation to transform discrete distance information into line segments with distinct feature is designed. Fit straight line segments by least square method considering sensor errors. Compute fitting error as criteria of threshold automatic adjustment. Computing Mahalanobis distance between lines in polar coordinates,achieve line merging to reduce storage contents of line features under the circumstance of retaining all the environment information. Comparing to the Split-and-Merge algorithm,solve the problem of sensitivity of threshold parameters. It is proved by experiments that the proposed algorithm can effectively detects line features of environment.