Autonomous Emergency Braking Strategy Based On Road Adhesive Coefficient Estimation

The intelligent and automated connected vehicles is a tremendous development trend in the automotive field today,and active safety systems such as AEB that can reduce traffic accidents have attracted widespread attention.AEB can acquire the information of the detected object based on visual sensors such as millimeter wave radar,and then make use of its algorithm module to avoid collision.The control module plays an important role of the AEB algorithm,in which the maximum deceleration of the vehicle is the key to safety distance.And this need estimating the road information such as friction coefficient.On the other hand,the AEB control system must conform to the driver's driving experience,so the driving habits of the driver need to be studied.For the AEB control system considering the estimation of the road adhesion coefficient,when having finished research papers at home and abroad,the visual sensors were studied and tested firstly.Then road slope and vehicle weight were estimated with directional forgetting recursive least squares algorithm and vehicle longitudinal dynamics.In order to obtain more accurate acceleration information,a more robust longitudinal dynamic model was identified using the lasso-based least square method and then acceleration was used to estimated vehicle speed using Kalman filter.By the way,the mathematical statistics method was used to find working time of online estimating algorithm based on scene segmentation.Finally,a multi-level warning anti-collision system is designed by combining the safety distance model and the collision time.The main content of this article can be summarized as follows:1)Select the sensor.After comparison of the current mainstream's sensing sensors such as lidar,millimeter wave radar,ultrasonic radar,and camera,millimeter wave radar was selected as the sensor to use.Then the millimeter-wave radars of various manufacturers in the market were compared,and then Delphi's 77 Ghz millimeter-wave radar was selected.And it was found that the design needs can be satisfied after testing.2)Identify vehicle longitudinal dynamic model.Firstly,the dynamic model of acceleration and deceleration was modeled theoretically,and then the directional forgetting recursive least square method was used to estimate vehicle weight and slope.Then the relevant factors were studied and a more robust longitudinal vehicle dynamic model was identified with lasso-based least square method.The mathematical statistics method was used to solve the problem of that the lateral and the longitudinal are coupled with.3)Estimate the road friction coefficient.The vehicle speed was estimated by using Kalman filter with acceleration which estimated by the longitudinal model.Then the slip was calculated with wheel speed which from tachometer.By analyzing the tire model,it was found that the effect of the tire model on different road surfaces in the small slip rate range was not shown,so the phenomenon of the tire contact zone was studied.It was found that the reason is that rolling resistance and free rolling pressure of tire model were ignored.Combined with the improved tire model,the road adhesion coefficient was estimated by using the force-slip ratio curve and the recursive least square method.By using the software Carsim,road friction coefficients were estimated under different roads.Simulation experiments show that the algorithm can estimate the road friction coefficients pretty well.4)Design of anti-collision system considering driver characteristics.The maximum deceleration was calculated by considering the road friction coefficient and the road slope.Then different safety distance models were studied and compared.By studying the characteristics of the driver,it is found that the driver is in favor of the multi-level warning system more than the single-level warning one.In addition,the driver's reaction time was modeled through data analysis.Finally,from considering the information of the collision time,the safety distance and the collision time were combined to design a multi-level warning anti-collision system.The simulation test results show that the multi-level warning system which takes into account both road information such as road friction coefficient and road slope,as well as driver characteristics can make the anti-collision system safer and more comfortable.