Vehicle-Non-Motor-Vehicle Traffic Conflict Automatic Recognition At Signalized Intersection Based On Vehicle Trajectory

Urban road intersections in our country gather traffic flows such as motor vehicles,non-motor vehicles,and pedestrians.This mixed traffic characteristic seriously affects the safety and efficiency of traffic operation.Non-motorized vehicles run complex in intersections,and their trajectories are highly random.They are prone to conflict with motor vehicles inside the intersection,and there are many types of conflicts,which have a greater negative impact on traffic safety and traffic efficiency in the intersection.Traditional signalized intersection safety analysis methods are mostly based on accident data for analysis and evaluation.However,this method has disadvantages such as difficulty in obtaining accident data,poor data quality,and waste of resources due to long observation periods,so its applicability is low.In recent years,traffic conflict technology has been widely used as an alternative evaluation method.Therefore,revealing the mechanism of the conflict between motor vehicles and non-motor vehicles at signalized intersections,proposing scientific methods for judging the conflict between vehicles and non-motors,and analyzing the severity of conflicts,and establishing an automatic judgment system for vehicle non-conflict is of great significance to solving the safety problems of intersections.Based on the Hunan Province graduate student research and innovation project,the thesis takes the signalized intersection safety evaluation as the starting point,and conducts in-depth exploration and research on the non-conflict problem of signalized intersections.Reveal the mechanism of machine non-conflict generation,analyze machine non-conflict characteristics,propose an improved TTC machine non-conflict discrimination model,analyze and verify the effectiveness of machine non-conflict discrimination methods,build machine non-conflict severity evaluation indicators,and establish machine non-conflict automatic discrimination models.Realize automatic discrimination of machine non-conflict.The main research content of the thesis can be divided into the following aspects:First,select typical signalized intersections in the city,use the video survey method,and use trajectory tracking software to extract the trajectory data of conflicting traffic flow between right-turning motor vehicles and straight non-motorized vehicles.Based on the traditional conflict discrimination definition,reduce hazards such as slowing down and changing lanes.The behavior and possible collision(the time from the collision point is less than 2 seconds)is based on the non-conflicting samples of the collection machine,and the tracker tracking software is used to extract the non-conflicting data of the machine.Secondly,study the operating characteristics of motor vehicles and non-motor vehicles at signalized intersections,and analyze the speed,deceleration,According to the characteristics of angle and distance to the collision point,according to the time when the vehicle reaches the conflict point and the avoidance measures taken,the avoidance behaviors are divided into four types:motor vehicle deceleration,non-motor vehicle deceleration,simultaneous deceleration and no avoidance Behavior,and analyze the characteristics of hedging behavior.Third,in view of the shortcomings of the current machine non-conflict discrimination method,an improved TTC discrimination method and conflict severity evaluation index are proposed,and the maximum time to collision(MTTC)and conflict time difference(TDTC)are selected as the evaluation index,and an improvement is proposed.Type TTC(Time to Collision)method of machine non-conflict discrimination method.Choose an example intersection in Changsha City,and adopt the improved TTC discriminant method,TTC discriminant method and post-intrusion time PET discriminant evaluation machine to be non-conflict.Compared with the TTC discriminant method,the accuracy rate is higher.Combining the probability of collision and the speed change value Delta-V before and after the collision,an evaluation index for the severity of aircraft-non-conflict conflict is proposed.The sensitivity analysis result of this indicator shows that the conflict index can reasonably explain the severity of the conflict.The non-conflict severity of a certain intersection in the city is graded.Fourth,obtain the trajectory sequence of motor vehicles and non-motor vehicles within the intersection range,build a machine non-conflict automatic discrimination model based on the improved TTC discrimination method and PET discrimination method,and implement python coding according to the machine non-conflict discrimination process,and develop machine non-conflict Automatic conflict discrimination system.The results show that the improved TTC automatic discrimination method and the manual detection discrimination method have the same discrimination results,and the conflict occurrence time is the same as the conflict location;the improved TTC discrimination method and the PET discrimination method have basically the same number of conflicts,but the conflict objects are obviously different.The machine non-conflict and discrimination system can obtain the conflict position of motor vehicles and non-motor vehicles,and obtain a heat map through visual analysis,which can provide reference for intersection managers.