A Real-time Construction of Raster Maps Based on Improved Data Normalization

Aiming at the current ORB-SLAM2 error problems in feature extraction and point cloud segmentation in the real-time acquisition of point cloud data scenarios, a real-time construction algorithm of raster maps based on data generalization is proposed to adapt to the real-time navigation and obstacle avoidance of the intelligent force and other task requirements. The algorithm considers the integrity of the environment configuration before verifying the real-time collection of point cloud data to guarantee the effect of real point cloud data, and then extracts the coordinate information of the point cloud data and improves the data generalization of the extracted point cloud data after retaining the three-dimensional coordinate information; then realizes the real-time construction of raster maps based on the point cloud data segmentation algorithm. The TUM open source dataset and the measured data are used for experimental verification to ensure the feasibility of the algorithm, and the real-time experiments are conducted with a monocular depth camera to realize the real-time construction of raster maps, which verifies the effectiveness of the proposed algorithm.