采用特征预处理ICP算法的机器人运动环境建图

针对机器人运动环境建图中迭代最近点(ICP)算法的扫描配准过程存在关联点对对应困难、迭代初值要求高的问题,首先,提出一种特征预处理的粗配准方法,以保证配准过程的迭代初值较小;然后,基于扫描点过滤思想对粗配准后存在较大误差的数据进行过滤,剔除测量噪声,提高点对关联准确率.结果表明:经过特征预处理的粗配准及噪声剔除后的改进ICP算法能够有效地进行扫描配准,解决机器人运动环境建图存在的问题.

Map Creation of Robot Motion Environment Using Feature Preprocessing ICP Algorithm

For the scanning registration process of iterative closest point(ICP)algorithm in map creation of robot motion environment, there are some problems, such as difficult correspondence of correlation point pair and high requirements for the iterative initial values, firstly, a coarse registration method with feature preprocessing is proposed, in order to ensure the iterative initial values of the registration process are relatively small. Then, based on the idea of scanning point filtering, the data with relatively large errors after coarse registration are filtered to eliminate measurement noise and improve the correlation accuracy of point pair. The results show that the improved ICP algorithm after coarse registration of feature preprocessing and noise elimination can effectively perform scanning registration, and solve the existing problems in map creation of robot motion environment.