Research On Active Safety Algorithms And Model For Intelligent Vehicle With A Laserscanner

Intelligent vehicles achieve driving safety mainly through perceptual system. Analysis of the data collected by on-board laserscanner provided the perceptual information of road environment based on improved algorithms of line extraction for the active safety model, to assist the intelligent vehicle in driving safety. The main work and achievements of this thesis are summarized as follows:With the collected data of the laserscanner, an improved algorithm SEF-LSM-BM for line extraction is proposed, which is based on the algorithms of successive edge following (SEF), least square method (LSM) and SEF-LSM. It combined the back-off mechanism, logical reasoning and feature merging to detect the road borders and obstacles, and identified drivable road region. The same real data scan was used in the experiments in order to evaluate different algorithms. Compared with algorithms of SEF, LSM and SEF-LSM, SEF-LSM-BM can detect the road borders, highlight the high-risk obstacles, and identify the drivable region rapidly and reliably.Based on the perpendicularity of vehicles, the algorithm PERP for line extraction and target detection of the front vehicle was used. Compared with algorithms of conics fitting, SEF and LSM, the processing speed of PERP is120-160times faster, with good matching and low square deviation. The distance, relative angle and velocity of the front target vehicle were measured by the millimeter-wave radar.Based on analysis and calculation of the physical quantities of each stage in the typical braking process, values of braking deceleration in different velocity were approximated, and the accuracy was above99.9%compared to the national standard, which verified the approximated driving model. After braking, the distance between ego-vehicle and the front vehicle was used as the key parameter to establish the active safety model. The safety degree was estimated respectively by the threshold of least time headway (LTH) and the critical safety distance (CSD). Compared to LTH, threshold CSD took the kinestates of ego intelligent vehicle and front target vehicle into account, which enabled the threshold to have better flexibility, and the active safety model better adaptability.