A Two-Dimensional Point Cloud Matching Method Based on ICP Improvement

The indoor map is the basis of indoor navigation and positioning. The core problem of indoor map construction by using two-dimensional lidar is to solve the matching problem between different point clouds. The standard ICP algorithm is easily affected by the initial position and interference points, so this paper proposes an improved algorithm based on ICP for 2D point cloud matching. This paper proposes a dynamic initial value matching method based on the highly correlated characteristics of the transformation parameters of adjacent point clouds. Then set the radius threshold to filter out the corresponding points to be matched to provide more ideal data input for the ICP algorithm. In this paper, the indoor parking lot with an area of about 2500 m² can be collected by using the vehicle 2D lidar equipment, which takes about 85 s to collect. And the global map can be obtained by using the improved ICP algorithm, which takes about 100 s to process the data. Experimental results show that this method has the remarkable characteristics of low cost and high efficiency. Compared with the total station measurement, the method can meet the requirements of the plane map with decimeter accuracy required by the navigation and positioning of the indoor parking lot, and it has a better matching effect in the practical application.