Font Size: a A A

Development Of Intersection Assistance Function And Research On Evaluation Methods Under Urban Conditions

Posted on:2022-04-30Degree:MasterType:Thesis
Country:ChinaCandidate:J B GaoFull Text:PDF
GTID:2492306332458664Subject:Vehicle Engineering
Abstract/Summary:PDF Full Text Request
With the acceleration of urbanization,as a transportation hub in urban working conditions,while giving full play to its flexibility,the intersection have gradually become the core of congestion and traffic accidents in urban road network.Serious traffic accidents at intersections have caused serious threats and damages to people’s property and life safety.In view of this situation,a novel auxiliary function of intersection based on trajectory prediction and conflict risk evaluation is developed in this paper,which mainly includes vehicle motion state estimation module,vehicle trajectory prediction module,collision risk assessment module and active braking module.In order to preprocess the data source mixed with noise,this paper first introduces the vehicle kinematics model based on constant acceleration model and constant turn rate and acceleration model respectively on straight and curved roads,and takes this model as the processing model of extended Kalman filter(EKF).The EKF is applied to integrate the signal of GPS and on-board vehicle sensors,and filter the noise data of both to realize the real-time estimation of vehicle position.Then,based on the road geometry information at the intersection,a cubic parabola model is established to simulate the driver’s future maneuvers.In the first step of trajectory prediction,the vehicle kinematics model and the cubic parabola model are fused and predicted by the exponential fusion weight function.The vehicle trajectory within 0-2s in the future could be predicted by taking the filtered vehicle motion parameters as input.In the second step,the trajectory prediction takes the Gaussian process model used for regression as the core model,and the trajectory points predicted in the first step as the training data of the model,and the vehicle trajectory in the future 2-2.5s and the uncertainty distribution in the prediction process can be deduced.Then,in order to facilitate the assessment of collision risk between vehicles and the timing of intervention when necessary,the two-dimensional geometry shape of vehicles is replaced by multiple circles.A conflict risk assessment model based on overlapping regions and a braking decision algorithm based on TTC are established.When there is a risk of collision,it mainly uses early warning to timely warn the driver to take over the vehicle so as to pass the intersection smoothly.However,when the driver does not make any action or brakes late,the active braking system will conduct braking intervention to avoid collision.Finally,in order to match the braking effect of the real vehicle and the simulation system,the ESC braking response characteristics of the real vehicle were studied,and the ESC braking response model was established by using feedforward control,feedback control and the first order inertia lag link.By referring to the vehicle-vehicle crash test scenarios at standard intersections provided by E-NCAP,this paper builds typical virtual intersection scenarios in SCANeR,such as left turn into path,left turn across path/lateral direction,Car-to-Car Front turn-across-path,left turn across path/right turn.Each sub-model and the integrated auxiliary model of the intersection are developed in Matlab/Simulink.Finally,a co-simulation platform based on SCANeR and Matlab/Simulink is built,and the co-simulation debugging and verification of the model are carried out.At the same time,the auxiliary function of the intersection is evaluated based on E-NCAP specification.The results show that the vehicle motion state estimation algorithm based on the EKF can obviously filter the noise in the signal,which provides a guarantee for the algorithm on the data source;The predicted trajectory obtained by the fusion trajectory prediction method is closer to the actual vehicle trajectory,which is better than that by using the kinematics method alone;The established ESC braking response model can well simulate the braking effect of real vehicle ESC and meet the operational requirements;The auxiliary function of the intersection developed in this paper can timely alert drivers and conduct robust braking in an emergency to avoid collisions.The effectiveness of the function is also verified from the’Good’ of the evaluation result.
Keywords/Search Tags:Intersection Assistance, Vehicle Trajectory Prediction, Extended Kalman Filter, Gaussian Process Regression, Collision Risk Assessment, E-NCAP Evaluation Method
PDF Full Text Request
Related items