Research On Laser Radar-based 3D Environmental Modeling For Unmaned Ground Vehicle

With the great development of laser radar scanning technology and the expansion of unmanned ground vehicle application, the applications of laser radar gradually spread to environmental perception, augmented reality, environmental modeling and many other fields of unmanned vehicle. Therefore, the research in the field of laser radar applications in unmanned ground vehicles has become an inevitable trend. Meanwhile, the research in the technology of processing laser radar data will definitely have a great significance on the development of the whole field of unmanned ground vehicle. Under the background of the research in 3D environment modeling for unmanned ground vehicles which is based on laser radar technology, this thesis mainly covers LIDAR Point Cloud based tele-presence method, adaptive matching algorithm of LIDAR Point Cloud Data and applications of LIDAR Point Cloud Data based 3D map modeling. By studying related technical method and carrying on verification experiment, further research in the technology of laser radar based 3D environmental modeling for unmaned ground vehicle has been done.The main contributions are shown below:1. A LIDAR Point Cloud based tele-presence method is proposed. First, the laser radar raw data should be processed by data filtering, terrain segmentation and other preprocess procedures. Then the 3D environmental model including surface obstacles and terrain could be established by the way of data interpolation and gridding process. This method has been proved to be effective through validating the measured data of typical environment.2. The adaptive ICP matching algorithm(ANICP) which is based on NDT algorithm is proposed. Through the process of pre-matching, fine-matching and the scene discrimination mechanism which is based on the analysis of the point cloud, ICP algorithm and NDT algorithm could be combined in a appropriate way. The verification experiment of adaptive matching point cloud data from different environments has been carried on. And the experimental result shows that this proposed algorithm can ensure the accuracy and meanwhile take the matching efficiency into account.3. The said algorithm has been applied in map modeling by the way of two frame matching and the whole group consistent position estimate. On the basis of ANICP algorithm in the elimination of positioning errors, the whole group consistent position estimate is introduced to the process of map modeling. Thus the cumulative error can be further suppressed by using this method. On this basis, a path planning analysis software has been developed with the function of planned path accessible analysis and energy assessment.Through the evaluation experiment of unmanned vehicle path planning in a typical environment, the effectiveness of the 3D point cloud model in the field of unmanned vehicle path planning has been proved.