Research On Decision Algoritm And Performance Test Method Of Lane Departure Warning System

With the global development of urbanization and extensive use of vehicles, transportation problem gets more and more people's attention and concern.Based on analysis of the causes of a large number of fatal accidents, lane departure accidents take up a large proportion due to driver's fatigue, inattention and other subjective factors. Accordingly, it is necessary to provide a system to alert the driver before lane departure happened, it can assist the driver to keep vehicles in correct driveway, reduce traffic accidents and enhance vehicle's active safety efficiently.Lane Departure Warning System (LDWS) is one of the popular research field of the intelligent driver assistance systems, generally referring access to the road front information by visual sensor, combined with vehicle speed and other related parameters of vehicle moving state and the set of warning time from the driver, to determine whether there is a trend of lane departure from the vehicle driving lane, if system judge the vehicle be in departure state, through the way of sound or vibration issue the lane departure warning to the driver, assist the driver to avoid or reduce the lane departure accidents.At present, many foreign auto corporations and research institutes have launched an extensive research on LDWS, some products have reached a commercial application stage. In our country the research of LDWS just confined to universities and research institutions, and there is no mature theoretical basis and test method, so it is necessary to carry out in-depth theoretical and applied research on LDWS.In this paper, combining with the research and development needs of the FAW Group project:"Development of monocular vision-based lane departure warning system", after researched the presented Decision Algorithm of LDWS, to improve the R & D capabilities, based on the laboratory's early LDWS Decision Algorithm and the theory of the Driver Optimal Preview Acceleration Model (DOPAM), around the calibration of system parameters, improvement of decision algorithm, establishment of simulation testing environment and research on performance test method, the work of this paper can be concluded as:(1) On the previous research on image processing and decision algorithm of the State Key Laboratory of Automobile Dynamic Simulation, this paper firstly achieved the decision algorithm based on the DOPAM in the test platform of the software.Since this warning decision algorithm required too many movement state parameters of the vehicle and can't be easily measured of vehicle test, at the same time, the algorithm can't meet the real-time requirements, through the estimation algorithm to get the parameters are very difficult and offen inaccurate, so this warning algorithm will not issue the alarm in time.In this paper, combined with the popular Time to Lane Crossing (TLC)decision algorithm and the need for engineering application,the paper proposed a Preview Time to Lane Crossing (P-TLC)decision algorithm to reflect the true characteristics of driver behavior. From the simulation results, the algorithm can effectively reduce the false alarm rate of the system and improve the efficiency and rationality of the system.(2) Vehicle joint simulation test environment is set up by using CarSim and Simulink software, in order to verify the efficiency and rationality of different decision algorithm and form a set of LDWS performance test method and parameter matching calibration.With their own car production plan, FAW Group can provide a clear design requirements and control parameter about selecting LDWS, and evaluate the product performance and technical indicator of the supplier.(3) On the fixed driving simulator platform of the laboratory, carried out the response time of test trials on the three typical drivers (comfort type, normal type, sport type), following carried out the vehicle test about the warning time parameter calibration of the LDWS. Then, based on the previous research on visual processing algorithms of the laboratory and the intelligent visual perception system of Japan's Kumamoto University, this paper carried out the installation of camera matching and parameters calibration work and the vehicle field tests. The results can be seen from the lane detection algorithm can effectively identify the traffic conditions ahead. Finally, this paper proposed a set of LDWS performance test method.