Research On Vehicle Safety Lane Change Warning Strategy Based On Vehicle-to-vehicle Collaboration

As a core part of the vehicle safety warning system,the vehicle safety lane change warning strategy analyzes the vehicle driving condition data obtained from the vehicle perception layer,using machine learning,pattern recognition and other methods to evaluate the danger degree of the lane change,giving the driver reasonableness.The early warning information of the vehicle and make the active safety system take over the manipulation of the vehicle when necessary,and even activate some passive safety functions in advance to ensure the safety of vehicle driving and collision,which is of great significance to the safety and reliability of the vehicle.Nowdays,it has been an important tendency in the automotive industry that such an integrated safety system that fully utilizes the advantages of active safety and passive safety and realizes realize active and passive safety integration.However,a series of technical difficulties in the current vehicle lane change warning strategy have not been well solved,such as how to quickly and effectively recognize the driver’s lane change intention,how to reasonably predict the trajectory of the lane change,how to provide effective early warning to drivers on the premise of ensuring a low false alarm rate.In view of the above-mentioned key technologies that need to be improved urgently,this paper proposes a vehicle safety lane change warning strategy based on vehicle-vehicle coordination,and performs related verification analysis.The specific research content is as follows:Firstly,this paper proposes a driver lane change intention recognition method based on PCA and GMM-HMM.A simulation driving platform is used to collect driver operation and vehicle operation information during lane change.Standardize the obtained data to eliminate the impact of each data due to different dimensions and value ranges.Based on the standardized data,a PCA analysis was performed.On the premise of fully preserving the amount of information in the original data,the first 3 principal components that can characterize the driver’s lane change intention were extracted for subsequent pattern recognition.A lane change intention recognition model based on GMM-HMM is established.The input of the model is the first 3 principal component sequences captured by the sliding time window method,and the output is the probability of each lane change intention.An offline model learning method was established using the EM algorithm,and model parameters were obtained using the training set.The test set was used to verify the accuracy and timeliness of the model recognition.The results show that the lane recognition intention recognition using this method can reach nearly 100% 1.5s after the lane change start point.Secondly,a vehicle lane change collision warning model based on lane change trajectory fitting is proposed.By analyzing the lane change data collected in the experiment,comparing the advantages and disadvantages of various lane change trajectory description methods,it is determined that the lane change trajectory is described by a five degree polynomial with a high degree of fit.A lane change sample database was established,and a lane change trajectory prediction model was established based on the sample database data and 5th-degree polynomial fitting.Starting from the prediction of vehicle motion trajectory,a rectangular vehicle model was introduced,and a calculation method of collision time TTC was established.Then the time to avoid collision TTA is analyzed and calculated.The relationship between TTC and TTA was comprehensively analyzed,and a risk coefficient R was established to quantify the danger of lane change collision.Finally,a lane change warning mechanism was designed based on the time line of the lane change process and based on the danger coefficient.Finally,the established lane warning safety warning system was tested and verified.The simulation driving platform used for lane change data collection has been expanded,and the established lane change safety early warning mechanism is extended to the software part of the platform.A typical vehicle trajectory is extracted from the NGSIM data as the motion state of the vehicle in the target lane.For the two dangerous scenarios in lane changing conditions: the target lane vehicle is behind the own vehicle and the target lane vehicle is in front of the own vehicle.The driver drives according to daily operation habits and obtains relevant data.The results of both experiments show that the lane change safety warning established in this paper can accurately and timely find the driver’s lane change intention,and judge the collision danger level through trajectory prediction,and give reasonable warning signals to ensure the lane change safety.