基于距离聚类的圆柱类实体路标特征提取算法

Feature Extraction of Circular Landmark Based Numerical Inference and Geometric Analysis

针对移动机器人激光扫描仪探测定位算法计算量大,实际应用受限,提出了基于距离聚类的圆柱类环境特征提取算法。通过数据预处理,滤除噪声干扰等无效数据;通过距离聚类的区域分割,将有用的激光测距数据分类为归属于各个环境路标特征点的数据子集,并去掉各类不合理特征点;通过几何计算进行参数拟合,提取出圆柱类实体路标的中心位置及其直径。仿真结果表明:该算法能够有效滤除噪声干扰,实现准确的路标特征提取。

Aiming at the problems that the detecting and location algorithms of laser range sensor on mobile robot had a large amount of calculation,which limited its practical application,an algorithm of circular type environment feature extraction algorithm based on distance cluster was presented.The noise and interference were filtered by data pretreatment,the useful data from laser range sensor were classified,and every landmark feature was extracted by segmentation based on distance cluster,and the unreasonable data points were removed.The centre of circular type landmarks and their diameters were extracted by parameter fitting of geometry calculation.The simulation showed that the algorithm could filter noise disturbance effectively,and could extract the feature of landmarks exactly.