LASER SLAM METHOD BASED ON SUBGRAPH MERGING AND POSE OPTIMIZATION

摘要

#$%^&*AU2020101833A420200924.pdf#####ABSTRACT The present invention discloses a laser SLAM method based on subgraph merging and pose optimization. The laser SLAM method includes the following steps: 1) acquiring 5 environmental information by laser radar scanning match, and mapping laser information to a world coordinate system by performing interframe match and coordinate transformation on laser data to generate a subgraph sequence; 2) optimizing pose information of a robot in the subgraph sequence generated in step 1); 3) adding the optimized subgraph and pose information into a candidate set, and scoring the candidate 0 set according to a standard to construct information tables; 4) screening the information tables in step 3) to obtain a compressed table, and then, merging a plurality of subgraphs together by using a data association method; 5) globally optimizing the plurality of subgraphs; and 6) generating a global grid map: continuously moving the robot to visit different places until a global map is generated. According to the present invention, the 5 global map is divided into the plurality of subgraphs by virtue of the idea of a divide-and-conquer method, and the global map is optimized by solving the problem of the subgraphs.1/4 Laser radar lsdlodectn initial dataClsdoodectn Frontend Scanning match - - Pose of robot feature extraction GassNei. 2Candidate set Point cloud match Sugah optuntalignment - Sugrap subgraph information Rear end -r Map merging Global map Fig. I Fig. 2