Correcting drifting error of mobile laser scanner using model-based loop closure detection

MLS (Mobile Laser Scanner) can efficiently obtain a point cloud of indoor GNSS-denied environment. However, it still suffers from inaccuracy due to the accumulated drifting error of simultaneous localization and mapping (SLAM). To overcome the above challenge, a novel method for correcting point cloud obtained from MLS using building information model (BIM) is proposed. First, ray casting is used to generate virtual scans based on the BIM model. Then, the key frame map is constructed by extracting features of virtual scans. Model-based loop closure detection (M-LCD) and six degree of freedom (6DOF) estimation are conducted to correct the drift error of MLS when loop closure is detected. The proposed solution is evaluated on popular benchmark datasets and the experiments results confirms that the accuracy of point cloud is significantly improved.