Research On Path Planning And Path Tracking For Vehicle Collision Avoidance By Steering Control Based On Dangerous Potential Field

With the increasing number of vehicle in China,the road traffic safety situation is becoming increasingly serious,especially the collision accident,so more and more people pay attention to traffic safety.But the current passive safety technology such as airbags and seat belts can not meet people’s need.Therefore,the manufacturers have developed varieties of pre-crash safety systems,such as Mercedes-Benz’s pre-safe system and Volvo’s CWAB(Collision Warning with Brake Support).These systems only provide early warning and avoid collisions by lowering the speed of the vehicle through brakingin when in the event of danger.However,when the distance between the car and the obstacle cannot meet the minimum safe braking distance,the accident cannot be avoided.Research shows that by actively steering or changing lanes,cars can complete the collision avoidance process in a shorter distance.Therefore,the collision avoidance path planning technology,which is based on active safety,can guide the vehicle to avoid collision,and this technology is helpful to improve the road traffic safety.Based on the analysis and summaries of the research status of active collision avoidance systems from domestic and abroad,aiming at the existing shortages of collision avoidance system in emergencies,this paper proposes the idea of avoiding collisions through planning steering path.The main research contents are as follows:After analyse the basic principle and defects of the traditional artificial potential field,a three-dimensional dangerous potential energy field model is used to plan the active steering collision avoidance path.In order to keep the vehicle in the middle of the main lane,the center line of the main lane is used as the attraction point to design the road potential field.A new obstacle repulsion potential field is designed by the ellipse distance in the function of the traditional artificial potential field,and combining the smoothness of the polynomial curve.Finally,increase the repulsive potential field of the road boundary to further restrict the range of motion of the vehicle during the steering collision avoidance process,to avoid the possibility of collision between the vehicle and the road boundary.In order to control the vehicle follows along the planned path,a three-degree-of-freedom vehicle dynamics model is established based on the theory of model predictive control,and after appropriate simplification,a predictive model is established with the front wheel steering angle as control variable.In order to reduce the amount of calculation of the model and satisfy the real-time requirement of the system,the nonlinear model is linearized and discretized into a linear time-varying model.The objective function is designed according to the requirement of the model controller to track the desired path quickly and smoothly,and the relaxationfactor is added to prevent the problem appearing of the unsolvable objective function.Considering the stability requirements of the vehicle in the course of path tracking by steering control,adding multiple constraints of the kinematics and dynamics for the vehicle,solving the optimization problem with constraints in the control period and realizing the vehicle collision avoidance path tracking by steering control.The monocular vision system is used to collect external environmental information,and the imaging model of the camera and the transformation relationship between the four coordinate systems are established.The camera parameters are calibrated using the plane template method after fixing the camera position and simplifying the processing,the theoretical formula of the inverse perspective transformation is derived.Finally,after preprocessing of the acquired image,the lane markings are detected using the Hough transform combined with the actual road background.Using the SURF algorithm to perform feature matching on the obstacle template image and the actual road image to detect obstacle information in the road.The experiment results showed that the image processing process in this paper can obtain external environmental information accurately.Finally,the Carsim/Simulink co-simulation platform is built and the simulation conditions under different conditions is set up.The simulation results showed that the dangerous potential field model can plan a safe and smooth steering collision avoidance path.The designed model predictive path tracking controller can control the vehicle to track the desired path quickly and steadily,and at the same time meet the multiple constraints of the kinematics and dynamics for the vehicle.