Research On Active Collision Avoidance System Based On Driving Risk Identification

The improvement in transportation convenience has caused a dramatic increase of cars,and at the same time,many traffic safety problems have arisen.People’s requirements for vehicle driving safety are increasing.In order to further reduce the driver’s driving burden and thus improve the driving safety,the research of automobile active collision avoidance system is very important.In this paper,a combination of theoretical algorithm improvement analysis and simulation verification is proposed,and an active collision avoidance system scheme based on driving risk identification is proposed.According to the key issues of the collision avoidance system,the focus is on the identification of the driving risk and the decision of the collision avoidance path.A series of studies have been carried out on control and other aspects.Firstly,from the information of natural driving events,the vehicle motion characteristics that can significantly characterize driving behavior are screened.Based on driver’s driving age and other related information,the improved LEC evaluation method is used to re-integrate the information and establish a driving risk assessment index set.The mapping relationships among vehicle status information,driver information is established,and driving risk levels.Based on the comprehensive consideration of driver and vehicle status,the K-means clustering algorithm is used to classify and identify sample data and analyze and study the clustering results.Secondly,considering the dynamics of the environment during the collision avoidance of the cars,an improved path planning algorithm is proposed to overcome the shortcomings of the local minimum and the unreachable target of the traditional artificial potential field method.According to the actual driving expectations,the gravitational potential field between the car and the road centerline and the repulsive potential field model based on the limit of the collision avoidance path are combined,and the relative motion of the vehicle and the sidecar during collision avoidance is combined to consider the relative speed and model of obstacle repulsive potential field with distance factor.The position coordinates are obtained by solving the car’s longitudinal balance(gravity and repulsion)equation,and the obstacle avoidance path is obtained through curve fitting.Based on comprehensive consideration of dynamic traffic environment information,vehicle motion status information(position,steering wheel angle,speed,etc.)and vehicle dynamic constraints,a driver ’s preview following model of lateral acceleration is established according to the driver ’s preview tracking theory to avoid Tracking control of the collision process.In order to improve the tracking accuracy of the path,a model predictive control algorithm is used to derive the design from the constraints and the controller objective function,and a path tracker based on the vehicle kinematics model is established.Finally,Matlab and Carsim software are used to verify the path planning and path tracking models.The simulation results of the path planning model show that the improved potential field method is better than the traditional artificial potential field method,and the planned path performance of the improved potential field method is more adaptable.The simulation results of the path tracking model show that both the path following model based on driver preview theory and the model based on model predictive control can completely track the collision avoidance path of the vehicle,but the model based on model predictive control is better than the model based on driver preview theory.The MPC tracking model of predictive control not only improves the tracking accuracy but also has a better tracking effect.