Low-cost fixed-angle ground-based lidar integration with point cloud registration

Abstract

Aiming at the problems of high cost and large size of the traditional mechanical lidar scanning point cloud, a point cloud data acquisition hardware system and a single-site cloud registration procedure were developed by using the prismatic lidar with low cost, small size and petal-shaped point cloud. Since the density of the point cloud collected by this lidar is time-dependent, in order to obtain a high-density point cloud, each station adopts a data collection method in which the motor-controlled lidar rotates 22.5 degrees each time, rotates 16 times, and scans the environment for one week. Using the self-developed station data processing programme, the data from each station were aligned according to the angle of the data by rotating the data through the space vector rotation algorithm. In the stage of inter-station point cloud registration, the original feature constraints of the multi-site cloud are obtained, and the error equations are derived from the constraints through the initial solution of all the station transformation parameters and unknown points except the control points. The weight function established by each constraint error is used as the constraint for iterative settlement until the iteration conditions are met, and all site space transformation parameters and location coordinates are output to achieve overall registration of multi-site cloud. This experiment shows that the point cloud data collected by the self-developed low-cost lidar has high density, high resolution, and the accuracy after registration is about 2cmin the nominal accuracy of prism lidar hardware, which has strong practicability and feasibility.