基于聚类的迭代双向最近点机器人位姿估计

An Iterative Dual Closest Point Method for Robot’s Pose Estimation Based on Clustering

  • 摘要: 针对迭代最近点(ICP)算法在存在严重遮挡的情况下容易陷入局部最小值的问题,对最近点规则(CP)进行了修改,提出双向最近点规则(DCP).DCP规则包含两次CP规则对应,使计算量增加了一倍.为了降低算法的复杂度,继而提出基于聚类的迭代双向最近点(IDCP BoC)算法.IDCP BoC对扫描数据进行聚类,在聚类的基础上进行数据精简.在相邻两次迭代的残差之差小于某个阈值之前,用精简数据进行迭代以提高计算速度,之后再改用非精简数据以保证精度.实验结果表明,IDCP BoC算法能够有效避免陷入局部最小值的问题且其实时性也是可接受的.

     

    Abstract: To overcome the problem of local extrema existing in iterative closest point(ICP) algorithm when severe occlusions occur,the closest point(CP) rule is modified and dual closest point(DCP) rule is proposed.DCP rule contains twice CP correspondences so that computation complexity is doubled.To decrease the computation complexity,iterative dual closest point based on clustering(IDCP BoC) is proposed.Scan range points are divided into clusters and then a procedure of data reduction is conducted.The reduced data set is used for iterative computation before the error of two consecutive iterations’ residual errors is less than a preset threshold to speed up the algorithm,and the data set without reduction is used after that to guarantee the accuracy.Experimental results show that IDCP BoC can avoid the problem of local extrema effectively and its real-time performance is also acceptable.

     

/

返回文章
返回