Rollover And Sideslip Recognition Method Of Front Vehicle And Decision Planning Method Of Emergency Collision Avoidance On Highway

The current researches of autonomous driving on highway mainly focus on the normal driving conditions and there are few researches of autonomous driving on highway when the front vehicle is in rollover or sideslip.In these occasions,the researches of autonomous driving are faced with the difficulties of recognizing the rollover and sideslip state of the front vehicle and guaranteeing the security of the autonomous vehicle.To address the above problems,this paper has researched the rollover and sideslip recognition of the front vehicle,and the decision planning method of emergency collision avoidance.To solve the problem that the autonomous vehicle cannot recognize the rollover state of the front vehicle effectively,a front vehicle rollover recognition method based on the characteristic angle of rollover is proposed.Firstly,the front vehicles are detected using a vehicle detection technique based on the shadow of a chassis.Some feature points of the vehicle regions in two adjacent images have been picked out using proposed adaptive threshold probability-guided random sample consensus algorithm.Subsequently,a new metric-characteristic angle of rollover is proposed,which can represent the changes of the roll angle of the front vehicle between two adjacent moments.Finally,a rollover recognition method using the historical characteristic angles of rollover to recognize the rollover state of the front vehicle is proposed.This makes it possible to recognize the rollover state of the front vehicle using the information of the forward camera of the autonomous vehicle.To solve the problem that the autonomous vehicle cannot recognize the sideslip state of the front vehicle effectively,a front vehicle sideslip recognition method based on the fusion of multiple sideslip features is proposed.Firstly,two key features of sideslip process are extracted.There is an extreme point on thecentrifugal acceleration curve during the sideslip process;The vehicle will slip out of the lane quickly after a sideslip occurs.Then,four sideslip characteristic quantities which can quantitatively describe the sideslip features are extracted,based on the perception information of the forward camera and radars of the autonomous vehicle.The four sideslip characteristic quantities arecentrifugal acceleration of front vehicle,speed of front vehicle which is perpendicular to the direction of the lane line,lateral distance from the edge of front vehicle to lane line,time from the current moment to the edge of front vehicle reaching the lane line.Finally,a sideslip recognition method using the fusion criterion of multiple sideslip features to recognize the sideslip state of the front vehicle is proposed,which makes it possible to recognize the sideslip state of the front vehicle using the perception information given by the forward camera and radars of the autonomous vehicle.To improve the safety of autonomous vehicle when the front vehicle is in rollover or sideslip conditions,a decision control strategy in these conditions is proposed.Firstly,a model of constant turn rate and acceleration(CTRA)was established to predict the trajectory of the front sideslip or rollover vehicles.Then,a feature vector of driving which only restricts the first derivative of trajectories is proposed to describe the driving process or collision avoidance process.Subsequently,a trajectory planning method based on the feature vector of driving is proposed,to plan the collision avoidance trajectories of autonomous vehicle.And a MPC trajectory tracking controller is designed to track the planned trajectories.In order to verify the validity of the proposed rollover,sideslip recognition methods,and the decision planning method,a cross-platform hardware-in-the-loop platform integrating d SPACE Simulator,video camera obscura and VTD is built and then the validation experiments are conducted.The results show that proposed rollover and sideslip recognition methods can recognize the rollover or sideslip state of the front vehicle on straights and corners in real-time,and the proposed trajectory planning method based on the feature vector of driving can deal with the conditions that the front vehicle is in rollover or sideslip states.The planning trajectories of the method can ensure that the autonomous vehicle effectively avoids collisions with the front rollover or sideslip vehicle,thus improve the safety of autonomous vehicle.