NDT-based robot positioning system for large scale diversity environment

High accuracy positioning is an important challenge for robots in large scale environment. In order to reduce the localization errors in the environments, an efficient Normal Distributions Transform (NDT) localization method with multi-sensor fusion data fusion, namely FPCR-NDT-localization, is proposed. Firstly, the laser point cloud is pre-processed to remove the ground point cloud and the laser feature point cloud is extracted to reduce the laser point cloud alignment. Secondly, the inertial measurement unit pre-integration results are used to estimate the LIDAR interframe state transfer volume, and the point cloud alignment efficiency is accelerated by the feature point-based FPCR-NDT alignment method. Finally, the IMU pre-integration results and the LIDAR inter-frame alignment are fused to estimate the robot pose in the global map. In the experiments, the localization performance of HDL-localization, NDT-localization and FPCR-NDT-localization systems are tested using MulRan dataset. The results show that FPCR-NDT-localization has higher localization accuracy in different scenes and higher real-time performance compared to the original algorithm.