Research On Obstacle Recognition Algorithm And Application Of Anti-collision System Under Urban Road Driving Conditions

As one of the advanced driver assistance technology, vehicle collision avoidance system can assist the driver, relieve the driver’s fatigue, reduce the possibility of traffic accident, and improve vehicle active safety performance, which makes it have a broad market prospect. Thus, the researches on vehicle collision avoidance system are mostly still in the demonstration and experimental stage in domestic. The lack of hardware systems at right price and the high rate of false alarm are all urgent to be solved. Related research will be necessary.Based on vehicle collision avoidance system under urban road driving condition as the application background, choosing the single line laser radar as the main sensor, according to the scanning characteristics of the radar, this paper has designed an adaptive vehicle obstacle detection and tracking algorithm in front of host vehicle, built simulation models to compare the Kalman tracking results of several kinds of maneuvering target models under different simulation conditions, and obtained the safety distance model adaptive to the experimental car based on the real vehicle test data. The specific research contents are as follows:(1) Through the comparison of some actual data of traffic accidents occurred in our country in recent years, the research significance of collision avoidance systems is raised. The research status of collision avoidance systems at home and abroad is compared and analyzed. And several methods of active collision avoidance system obstacle recognition are introduced.(2) The whole process of anti-collision system obstacle recognition algorithm is made. According to the working principle of the automotive anti-collision system, the whole algorithm is divided into four parts, traffic information detection, obstacle recognition, target tracking and safety assessment. And the key technology of the algorithm is analyzed. Compared some kinds of traffic information sensor of anti-collision system, a laser radar is selected. Then the composition of the control unit and interface circuits are introduced, which lays the foundation for detailed Introduction of the algorithm.(3) The obstacle recognition algorithm with adaptive threshold is determined. The principle, advantages and disadvantages of several commonly used clustering analysis algorithm are introduced. Considering the scanning characteristics of the laser radar, the nearest neighbor clustering algorithm is adopted for obstacle recognition. The clustering threshold will be changed adaptively, according to the change of scanning distance, to enhance the accuracy of the algorithm.(4) Selection of maneuvering target model to study the adaptive Kalman filtering algorithm. The system principle and application of constant velocity model, constant acceleration model, Singer model and the current statistical model are compared. Taking the real urban road driving situation into condition, the current statistical model is selected as the maneuvering target model. And the adaptive Kalman filtering algorithm based on the current statistical model is used for target tracking. The influence of system parameters on the adaptive tracking algorithm will be analyzed.(5) Simulation analysis and real vehicle test. Simulink simulation models are built. The simulation results of constant velocity model, constant acceleration model and the current statistical model under the uniform motion, uniform acceleration motion and variable acceleration motion simulation conditions are compared and analyzed. And the real vehicle test platform is built for vehicle tests of clustering algorithm and target tracking algorithm. Safety assessment tests are carried out based on the driver preview safety distance model. The braking distance measured is dealt with through regression analysis so that a safety distance model adapted to the test vehicle’s collision avoidance system is obtained.