Research On LADAR Based Road And Obstacles Detection

As one kind of intelligent robots, autonomous land vehicle (ALV) has great research value and broad application prospects in the national defense, civilian traffic and scientific research fields. This dissertation carried out the research on laser range finder(LADAR) methods and technologies applied in ALV environment understanding theoretically and practically.The main contributions and innovations are as following:1) With plane-scanning model of ladar and extended kalman filter algorithm, a road-surface detection algorithm was improved. The classification of the ladar data was realized and the ladar data from the road was successfully extracted. Based on sequence ladar data, a method of extracting road regions was proposed, which solved the problem of ladar data losing.2) With adaptive threshold nearest neighbor clustering algorithm and obstacle detection criteria based on the height information, the obstacle detection was completed.3) Based on expert system and ensemble learning, a method of expert-integrated road and obstacles detection was proposed, and obstacles detection expert and road-surface extraction expert was designed. The results of road and obstacles detection were presented, and road and obstacles detection based on ladar was realized.4) With detachable-step ladar calibration algorithm, the ladar calibration was realized. The extrinsic calibration of ladar and camera was improved, and simulated annealing method was adopted on the initial results to optimize the calibration matrix, which not only reduced the calibration workload but also improved the calibration accuracy.Experiments indicated that the proposed method can not only detecte road but also classify and detecte obstacles successfully.