A Vehicle Trajectory Repair Method for Signalized Intersections Based on Roadside LiDAR

Abstract

The present invention belongs to the technical field of traffic target recognition, and particularly relates to a method for repairing vehicle trajectories at signalized intersections based on roadside lidar. The purpose of the present invention is to solve the problems of incorrect, missing, and discontinuous vehicle trajectory tracking caused by the existing vehicle trajectory prediction methods not considering signal phase changes, whether there are vehicles ahead, and the motion state of the preceding vehicle during the operation of vehicles at signalized intersections. The solution of the present invention is mainly based on the roadside lidar point cloud data, considering factors such as signal lights and vehicle queues, and using vehicle following models and motion models to repair the trajectories of vehicles with interrupted trajectories at different positions within signalized intersections, so as to improve the accuracy and continuity of trajectory data, thereby providing high-precision and complete traffic flow trajectory data for the holographic perception of road traffic flow, refined traffic signal control, and accompanied traffic travel guidance.

Description

Technical Field

[0001] The present invention belongs to the technical field of traffic target recognition, and particularly relates to a method for repairing vehicle trajectories at signalized intersections based on roadside lidar. Background Art

[0002] In recent years, with the rapid development of the concept of vehicle-road cooperation, traffic sensors have become one of the research hotspots in the fields of urban traffic and safety. As an important part of intelligent transportation, traffic sensors realize functions such as intelligent traffic signal control, traffic congestion prediction, and traffic safety reminders through information sharing between vehicles and road equipment, thereby improving the efficiency and safety of urban traffic, and further promoting the development of smart cities and enhancing the intelligent level of cities.

[0003] Currently, commonly used traffic sensors include video, coils, Bluetooth, microwave radars, etc., which mainly obtain macroscopic traffic parameters such as vehicle speed, occupancy, and flow. In comparison, the advantage of roadside lidar is that it can accurately measure various parameters such as the position, speed, and direction of vehicles, so it can more accurately identify and track vehicles, with higher safety and reliability. At the same time, lidar can also work normally under complex road and bad weather conditions, with a wider scope of application. These advantages make traffic control technologies based on roadside lidar have broad application prospects in the traffic field.

[0004] Specifically, by setting up lidar and corresponding algorithm systems around intersections, the position, speed, direction, etc. of vehicles are monitored in real time, so as to accurately control the trajectories and operating states of vehicles. However, due to the inevitable problem of vehicle occlusion within the intersection range, the extracted high-resolution trajectory data is incomplete. Currently, some researchers have developed methods for repairing trajectories on sections or in highway scenarios. However, due to the existence of factors such as signal lights and queues, these methods cannot be directly applied to the signal control intersection scenario, so it is very challenging to repair vehicle trajectory data within the detection range of intersections. Summary of the Invention

[0005] The present invention proposes a method for repairing vehicle trajectories at signalized intersections based on roadside lidar, aiming to solve the problems of incorrect, missing, and discontinuous vehicle trajectory tracking caused by the existing vehicle trajectory prediction methods not considering signal phase changes, the presence of vehicles ahead, and the motion states of the preceding vehicles during the operation of vehicles within signalized intersections. The solution of the present invention is mainly based on the roadside lidar point cloud data, considering factors such as signal lights and vehicle queues, and using vehicle following models and motion models to repair the trajectories of vehicles with interrupted trajectories at different positions within signalized intersections, so as to improve the accuracy and continuity of trajectory data, thereby providing high-precision and complete traffic flow trajectory data for holographic perception of road traffic flow, refined traffic signal control, and accompanied traffic travel guidance.

[0006] A method for repairing vehicle trajectories at signalized intersections based on roadside lidar proposed by the present invention mainly includes three parts: vehicle trajectory data collection at signalized intersections, interrupted trajectory identification, and interrupted trajectory repair, as Figure 1 shown.

[0007] (1) Vehicle trajectory data collection at signalized intersections

[0008] Lidar devices are deployed on the roadside of signal-controlled intersections to collect vehicle data; the collected data is processed through background filtering, clustering, vehicle detection, and trajectory tracking to obtain the original vehicle trajectory data, and the trajectory data includes the position information of each vehicle at each moment; the detection range of the intersection is gridified to obtain more information about the trajectory, such as the approach lane and lane information.

[0009] (2) Interrupted trajectory identification

[0010] In the lidar scan data, interrupted trajectories caused by occlusion of target objects or outside the sensor's field of view are detected and identified; according to the trajectory data, it is judged whether the trajectory is an interrupted trajectory based on the relationship between the start and end positions of the trajectory and the detection range of the intersection.

[0011] (3) Interrupted trajectory repair

[0012] After obtaining the interrupted trajectories, they need to be repaired in chronological order. To repair the trajectories, the information of the interrupted trajectories, such as the interruption time and interruption position, is used to judge the state of the current traffic phase, whether the vehicle has passed the stop line, whether there are vehicles ahead, and if there is a preceding vehicle, whether its motion condition is an idle condition. Based on these judgments, corresponding trajectory repair schemes can be formulated, as Figure 2 shown. Description of the drawings

[0013] Figure 1 : The functional structure diagram of the present invention;

[0014] Figure 2 : Flow chart of trajectory repair solution;

[0015] Figure 3 : Schematic diagram of intersection grid. Specific implementation manner

[0016] A vehicle trajectory repair method for signalized intersections based on roadside lidar according to the present invention is based on roadside lidar point cloud data, considers factors such as signal lights and vehicle queues, and uses a car-following model and a motion model to complete the repair of vehicle trajectories with trajectory interruptions at different positions within a signal-controlled intersection, so as to improve the accuracy and continuity of trajectory data. The specific implementation steps are as follows:

[0017] Step 1.1: Install lidar on the roadside of a signal-controlled intersection to collect vehicle data; the original vehicle trajectory data can be obtained through background filtering, clustering, vehicle detection, and tracking processes;

[0018] The trajectory data of each vehicle contains x t , y t , t represents the moment, x t , y t represents the vehicle coordinates at moment t. Taking the lidar as the coordinate origin, the acquisition range is x t ∈[x min , x max , y t ∈[y min , y max ;

[0019] x min = -30

[0020] y min = -30

[0021] x max = 30

[0022] y max = 30

[0023] Where: x min , x max are the minimum and maximum abscissas of the vehicle; y min , y max are the minimum and maximum ordinates of the vehicle;

[0024] Step 1.2: According to the coordinate information of each vehicle, the speed v x,t , v y,t , v t and the acceleration a t can be calculated;

[0025]

[0026]

[0027]

[0028]

[0029] where: x t , x t+1 , y t , y t+1 are the x and y coordinates of the vehicle at times t and t + 1 respectively; v x,t , v y,t is the velocity in the x and y axis directions at time t; v t , v t+1 is the velocity at times t and t + 1; a t is the acceleration at time t;

[0030] Step 1.3: Grid the collection range at the intersection, divide it into row × col grid numbers, as Figure 3 shown;

[0031]

[0032]

[0033] where: row and col are the grid numbers in the x and y axis directions respectively, and can be set as row = 20, col = 20; length row , length col are the width and length of the grid respectively. In the case of row = 20, col = 20, length row = length col = 3;

[0034] Therefore, the grid number where each trajectory point is located can be determined, and the calculation formula for the grid number is as follows:

[0035]

[0036] where: grid is the grid number where the trajectory point (x t , y t ) is located;

[0037] Each imported stop line can be determined by two coordinates. Based on the position coordinates of the stop lines at the four entrances of the signalized intersection, the grid numbers where the stop lines at the four entrances are located can be determined. And according to the fact that the grid numbers where the stop lines at the north-south entrances change at an interval of 1, while those at the east-west entrances change at an interval of 20, and the grid number at the north entrance is greater than that at the south entrance, and the grid number at the east entrance is greater than that at the west entrance; changing according to this rule, finally, the corresponding relationship between the stop line coordinates at the four entrances and the approach roads can be obtained.

[0038]

[0039] Where: x p1,d , y p1,d , x p2,d , y p2,d represent the stop line coordinates of the approach road entry d , d represents the numbers of the four approach roads; when d = 0, 1, 2, 3, entry d are north, south, east, and west respectively;

[0040] Step 1.4: According to the grid number change rule, based on the stop line grid number information of the four entrances, the physical functional area Ω p of the intersection, the upstream functional area Ω u and the downstream functional area Ω d can be obtained. The upstream functional area Ω u and the downstream functional area Ω d of the intersection can be divided into lanes, and the lane number is lane, lane ∈ (1, n), where n is the sum of the numbers of upstream approach roads and downstream exit roads in the four directions;

[0041] Step 1.5: Based on the above calculation process, the trajectory data Α of all vehicles during the acquisition time period is statistically collected. A includes complete vehicle trajectories and interrupted vehicle trajectories;

[0042] The trajectory data of each vehicle in Α consists of multiple pieces of data, and each piece of data contains 11 data items, which are respectively:

[0043] [x t , y t , v x,t , v y,t , v t , a t , t, entry d , lane, id, grid]

[0044] Step2: According to the coordinates of the start and end points of each trajectory, determine whether it is an interrupted trajectory, and the set of interrupted trajectories is I;

[0045]

[0046] where: x start , y start , x end , y end are the starting and ending coordinates of the trajectory id, and δ id = 1 indicates that the trajectory id is a complete trajectory, and 0 indicates an interrupted trajectory;

[0047] The set I of interrupted trajectories is the set of all interrupted trajectory ids;

[0048] Step3: Determine the trajectory repair plan:

[0049] Step 3.1: Extract the trajectories in the set I of interrupted trajectories in sequence, and count the basic data of the interrupted trajectories one by one. It includes the vehicle position where the trajectory is interrupted, the interruption time, the lane where the trajectory is interrupted, the approach information, etc., and the distance from the position where the interrupted trajectory is interrupted to the stop line;

[0050]

[0051]

[0052] where: f t , f lane , f id , f grid are the vehicle coordinates, speed, acceleration, interruption time, approach, lane, vehicle number, and grid index where the interruption occurs, respectively;

[0053] In addition, the distance from the stop line when the trajectory is interrupted can be represented by f L , and the calculation process is as follows:

[0054] f L = L

[0055] The stop line coordinates of the approach entry d are x p1,d , y p1,d , x p2,d , y p2,d , and the stop line can be linearly represented by a straight-line equation,

[0056]

[0057] b = y p1,d - k × x p1,d

[0058]

[0059]

[0060] Where: k and b are the slope and intercept of the straight-line equation of the stop line respectively; l is the length (greater than 0) from the coordinate of the end point of the interrupted trajectory to the stop line; L is the vector distance with direction. If the coordinate of the end point of the interrupted trajectory does not pass through the stop line, L is negative; otherwise, L is positive.

[0061] Step 3.2: Based on the length f from the coordinate of the end point of the interrupted trajectory obtained in 3.1 to the stop line L , it can be determined whether f L is greater than 0. If so, go to Step 3.3; otherwise, go to Step 3.6.

[0062] Step 3.3: According to the lane where the interruption moment f t of the interrupted trajectory is located lane , determine whether there is a vehicle in front of the interrupted trajectory on this lane. If there is a vehicle, go to Step 3.4; otherwise, go to Step 3.5.

[0063] In all trajectory data Α, finding the trajectory data of the vehicle in front needs to meet the following processes:

[0064] First, determine whether there is a vehicle on the current lane:

[0065]

[0066] Where: vehicle = 0 indicates that there is no vehicle on the current lane; vehicle = 1 indicates that there is a vehicle on the current lane, but it does not necessarily mean it is the vehicle in front.

[0067] Therefore, extract the trajectory data that meets the conditions and calculate further. According to the calculation process in Step 3.1, the distance f from the trajectory data of other vehicles on the lane to the stop line can be extracted L ;

[0068] Set the initial value diff = 100:

[0069] diff dist = l L - f L

[0070]

[0071]

[0072] Where: diff dist is the distance difference between the interrupted trajectory at the moment f t and other vehicles on this lane. If the condition is met, assign diff dist to diff, and finally diff is ft The distance difference between the interrupted trajectory and the vehicle ahead; leading = 0 indicates that there is no vehicle in front of the interrupted trajectory; leading = 1 indicates that there is a vehicle in front of the interrupted trajectory;

[0073] According to the above steps, the trajectory data of the vehicle ahead at the interruption moment f t can be extracted;

[0074] Step 3.4: Based on the trajectory data of the vehicle ahead and the interrupted trajectory data, predict the relevant data of the interrupted trajectory at the next moment based on the Intelligent Driver Model (IDM); and according to the predicted data, return to Step 3.3;

[0075] According to all the trajectory data Α, the parameters in the IDM model can be determined: the desired speed v s , the safe headway time T, the maximum acceleration a max , the comfortable deceleration b, the acceleration exponent δ, the minimum spacing s0, the time interval Δt; according to the end coordinates and speed of the interrupted trajectory, the desired speeds v x,s , v y,s , the maximum acceleration a x,max , a y,max , the comfortable deceleration b x , b y and the minimum spacing s x , s y ;

[0076] Specifically, the calculation process of the IDM model is as follows:

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083] In the formula: dvx is the speed difference in the x-axis direction between the interrupted trajectory and the vehicle ahead; are the speeds in the x-axis direction of the interrupted trajectory and the vehicle ahead at the moment f t respectively; dx is the absolute value of the interval in the x-axis direction between the interrupted trajectory and the vehicle ahead; are the positions in the x-axis direction of the interrupted trajectory and the vehicle ahead respectively; L vehicle is the vehicle length; gap xis the desired spacing in the x-axis direction in the current state; are the acceleration, velocity, and position in the x-axis direction of the interrupted trajectory at the next moment; Δt is the time interval;

[0084] Similarly, it can be calculated that the resultant velocity at the next moment of the interrupted trajectory can be calculated therefrom and the acceleration

[0085]

[0086]

[0087] f t = f t + 1

[0088] Other information f id , f lane , etc. can be obtained according to Steps 1.3 and 1.4;

[0089] Step 3.5: If there is no vehicle in front of the interrupted trajectory, then predict the relevant data of the interrupted trajectory at the next moment according to the variable acceleration motion model; and judge whether the predicted position exceeds the research range. If it exceeds the research range, end the prediction. Otherwise, continue to predict with the variable acceleration motion model until it exceeds the research range;

[0090] The parameters of the variable acceleration motion model are the maximum acceleration a max , the velocity response coefficient σ, and the time interval Δt. According to the maximum acceleration a max and the velocity at the moment f t of the trajectory interruption the maximum accelerations a x,max , a y,max in the x and y axis directions can be determined by solving the system of equations; the calculation process is as follows;

[0091]

[0092] There are two solutions to this system of equations. Select the solution with the same sign as the corresponding component velocity as the component accelerations a x,max , a y,max ;

[0093] The specific process of the variable acceleration motion model is as follows:

[0094]

[0095]

[0096]

[0097]

[0098] Where: Δx is the traveling distance;

[0099] Similarly, is consistent with the above calculation process;

[0100]

[0101]

[0102] f t = f t + 1

[0103] Other information f id , f lane , etc. can be obtained according to Step 1.3 and 1.4;

[0104] Step 3.6: If the interrupted trajectory does not cross the stop line, it is necessary to determine whether there is a vehicle in front of the interrupted trajectory, and the judgment criterion is based on Step 3.3. If there is a vehicle in front, go to Step 3.7; if not, go to Step 3.13;

[0105] Step 3.7: There is a vehicle in front of the interrupted trajectory. Determine the current signal phase. If the moment f t of the trajectory interruption is in the green light phase, go to Step 3.8; otherwise, go to Step 3.9;

[0106] Based on f t , and the intersection signal cycle, the current signal phase can be determined;

[0107] Step 3.8: There is a vehicle in front of the interrupted trajectory and the current phase is the green light phase. Then predict the relevant information of the interrupted trajectory at the next moment according to the IDM model; and according to the predicted trajectory data, return to Step 3.6;

[0108] Step 3.9: There is a vehicle in front of the interrupted trajectory and the current phase is the red light phase. Determine whether the vehicle in front is in the idle condition; if it is in the idle condition, go to Step 3.10; otherwise, go to Step 3.8;

[0109] Determine whether the vehicle is in the idle condition:

[0110]

[0111] where: mode represents the driving condition of the vehicle ahead; mod = -1 indicates that the vehicle ahead is in a non-idling condition; mode = -1 indicates that the vehicle ahead is in an idling condition; represents the acceleration of the vehicle ahead at time f t , with the unit of m / s; represents the speed of the vehicle ahead at time f t , with the unit of km / h;

[0112] Step 3.10: When the vehicle ahead is in an idling condition, calculate the position of the trajectory interruption time f t and the distance dist from the vehicle ahead; if the distance dist is less than the deceleration safety distance threshold d safe , then go to Step 3.11, otherwise go to Step 3.12;

[0113] The formula for the distance between the two vehicles is:

[0114] d f,l = l L - f L

[0115] where: d f,l is the distance between the interrupted trajectory and the vehicle ahead (greater than 0); l L is the distance from the vehicle ahead to the stop line, and the calculation process is the same as in Step 2.2;

[0116] dist = d f,l - L vehicle - s safe

[0117] The acceleration at the next moment of the interrupted trajectory is:

[0118]

[0119] where: is the speed of the interrupted trajectory at the interruption time f t ; L vehicle is the vehicle length; s safe is the minimum safety distance; the acceleration at the next moment of the interrupted trajectory;

[0120] According to the speeds along the x and y axes of the trajectory interruption time f t and as well as the corresponding acceleration can be calculated

[0121]

[0122]

[0123]

[0124] It can be obtained according to the above calculation principle;

[0125] Step 3.11: After the interrupted trajectory decelerates and stops waiting in the next moment; judge the current phase state; if it is a green light phase, return to Step 3.6, otherwise continue to maintain the stop waiting state until it becomes a green light phase;

[0126] Step 3.12: After the interrupted trajectory decelerates, return to Step 3.9;

[0127] Step 3.13: There is no vehicle in front, judge the phase t at the moment f when the trajectory is interrupted; if it is a green light phase, go to Step 3.14, otherwise go to Step 3.15;

[0128] Step 3.14: Predict the relevant information of the next moment of the interrupted trajectory with a variable acceleration motion model; calculate the distance between the predicted position and the stop line; if the distance is greater than 0, go to Step 3.3, otherwise go to Step 3.13;

[0129] Step 3.15: Calculate the distance dist between the position where the interrupted trajectory is interrupted at moment f t and the stop line; if the distance dist is less than the deceleration safety distance threshold d safe , then go to Step 3.16, otherwise go to Step 3.17;

[0130] Step 3.16: The interrupted trajectory decelerates to a stop; determine the current phase; if it is a green light phase, go to Step 3.3, otherwise continue to stop until it becomes a green light phase;

[0131] Step 3.17: The interrupted trajectory decelerates, judge whether the current phase is a green light phase, if it is a green light phase, go to Step 3.3, otherwise go to Step 3.15.

Claims

1. A method for repairing vehicle trajectories at signalized intersections based on roadside lidar, characterized in that: Step 1: Collect vehicle trajectory data at signalized intersections; Step 2: Identify interrupted trajectories; Step 3: Repair interrupted trajectories: Step 3.1: Sequentially extract the trajectories in the interrupted trajectory set I, and statistically analyze the basic data of the interrupted trajectories one by one; Step 3.2: Based on the length f of the end point coordinate of the interruption trajectory obtained in 3.1 from the stop line L , f can be determined L whether it is greater than 0; if so, go to Step 3.3, otherwise go to Step 3.6; Step 3.3: According to the lane f where the interruption trajectory interrupts at moment f t where it is located lane , determine whether there is a vehicle in front of the interruption trajectory on this lane. If there is a vehicle, go to Step 3.4; otherwise, go to Step 3.5; Step 3.4: Based on the leading vehicle trajectory data and the interrupted trajectory data, predict the relevant data of the interrupted trajectory at the next moment based on the Intelligent Driver Model (IDM); and according to the predicted data, return to Step 3.3; Step 3.5: If there is no vehicle in front of the interrupted trajectory, predict the relevant data of the interrupted trajectory at the next moment according to the variable acceleration motion model; and determine whether the predicted position exceeds the research range. If it exceeds the research range, end the prediction, otherwise continue to predict with the variable acceleration motion model until it exceeds the research range; Step 3.6: If the interrupted trajectory has not passed the stop line, it is necessary to determine whether there is a vehicle in front of the interrupted trajectory, and the judgment criterion is based on Step 3.3; if there is a vehicle in front, go to Step 3.7, if not, go to Step 3.13; Step 3.7: There is a vehicle in front of the interrupted trajectory, determine the current signal phase; if the moment f of trajectory interruption t is in the green light phase, go to Step 3.8, otherwise go to Step 3.9; Step 3.8: If there is a vehicle in front of the interrupted trajectory and the current phase is the green light phase, predict the relevant information of the interrupted trajectory at the next moment according to the IDM model; and according to the predicted trajectory data, return to Step 3.6; Step 3.9: If there is a vehicle in front of the interrupted trajectory and the current phase is the red light phase, determine whether the vehicle in front is in an idling condition; if it is in an idling condition, go to Step 3.10, otherwise go to Step 3.8; Step 3.10: The vehicle ahead is in an idle condition, calculate the position at the moment f when the trajectory is interrupted and the distance dist from the vehicle ahead; if the distance dist is less than the deceleration safety distance threshold d t , then go to Step 3.11, otherwise go to Step 3.12; safe ​ Step 3.11: The interrupted trajectory decelerates and stops waiting in the next moment; determine the current phase state; if it is the green light phase, return to Step 3.6, otherwise continue to maintain the waiting state until it becomes the green light phase; Step 3.12: After the interrupted trajectory decelerates, return to Step 3.9; Step 3.13: There is no vehicle ahead, determine the moment f when the trajectory is interrupted t and the phase it is in; if it is a green light phase, go to Step 3.14, otherwise go to Step 3.15; Step 3.14: Predict the relevant information of the interrupted trajectory at the next moment according to the variable acceleration motion model; calculate the distance between the predicted position and the stop line; if the distance is greater than 0, go to Step 3.3, otherwise go to Step 3.13; Step 3.15: Calculate the interruption moment f of the interruption trajectory t The distance dist between the position where it is located and the stop line; if the distance dist is less than the deceleration safety distance threshold d safe , then go to Step 3.16, otherwise go to Step 3.17; Step 3.16: The interrupted trajectory decelerates to a stop; determine the current phase; if it is the green light phase, go to Step 3.3, otherwise continue to maintain the stop until it becomes the green light phase; Step 3.17: The interrupted trajectory decelerates, and determine whether the current phase is the green light phase. If it is the green light phase, go to Step 3.3, otherwise go to Step 3.

15.

2. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 1, characterized in that: Step 1 includes: Step 1.1: Install lidar on the roadside of the signal-controlled intersection to collect vehicle data; the original vehicle trajectory data can be obtained through background filtering, clustering, vehicle detection, and tracking processes; The trajectory data of each vehicle includes x t , y t , where t represents the time, and x t , y t represent the vehicle coordinates at time t. With the lidar as the coordinate origin, the acquisition range is x t ∈ [x min , x max , y t ∈ [y min , y max ; x min ＝-30 y min ＝-30 x max ＝30 y max ＝30 where: x min , x max is the minimum and maximum abscissa of the vehicle; y min , y max is the minimum and maximum ordinate of the vehicle; Step 1.2: According to the coordinate information of each vehicle, the speed v of the vehicle can be calculated x,t , v y,t , v t and the acceleration a t ; where: x t , x t+1 , y t , y t+1 are the x and y coordinates of the vehicle at times t and t + 1 respectively; v x,t , v y,t is the velocity in the x and y axis directions at time t; v t , v t+1 are the velocities at times t and t + 1 respectively; a t is the acceleration at time t; Step 1.3: Grid the acquisition range of the intersection and divide it into row×col grid numbers; where: row and col are the numbers of grids along the x-axis and y-axis respectively, and let row = 20, col = 20; length row , length col are the width and length of the grid respectively. In the case of row = 20, col = 20, length row = length col = 3; Each trajectory point can determine its grid number, and the calculation formula for the grid number is as follows: where: grid is the grid number where the trajectory point (x t , y t ) is located; The stop line of each approach can be determined by two coordinates. Based on the position coordinates of the stop lines of the four approaches of the signalized intersection, the grid numbers where the stop lines of the four approaches are located can be determined; and according to the grid numbers where the stop lines of the north-south approaches change at an interval of 1, while the east-west approaches change at an interval of 20, and the grid number of the north approach is greater than the grid number where the stop line of the south approach is located, and the grid number where the stop line of the east approach is located is greater than the grid number of the stop line of the west approach; changing according to this rule, finally, the corresponding relationship between the stop line coordinates of the four approaches and the approach lanes can be obtained; where: x p1,d , y p1,d , x p2,d , y p2,d represent the coordinates of the stop line of the entry lane d , d represents the numbers of the four entry lanes; when d = 0, 1, 2, 3, entry d are respectively north-south, east-west; Step 1.4: According to the variation law of grid numbers, based on the grid number information of the stop line grids at the four inlets, the physical functional area Ω of the intersection can be obtained. p , the upstream functional area Ω u and the downstream functional area Ω d . The upstream functional area Ω u and the downstream functional area Ω d of the intersection can be divided into lanes, and the lane numbers are lane, where lane ∈ (1, n), and n is the sum of the number of upstream approach lanes and the number of downstream exit lanes in the four directions. Step 1.5: Based on the above calculation process, count the trajectory data Α of all vehicles during the acquisition time period. A includes complete vehicle trajectories and interrupted vehicle trajectories; The trajectory data of each vehicle in Α consists of multiple pieces of data, and each piece of data contains 11 data items, namely: [x t , y t , v x,t , v y,t , v t , a t , t, entry d , lane, id, grid].

3. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 2, characterized in that: Step 2: According to the coordinates of the start and end points of each trajectory, judge whether it is an interrupted trajectory, and the set of interrupted trajectories is I; where: x start , y start , x end , y end are the starting and ending coordinates of the trajectory id, and δ id = 1 indicates that the trajectory id is a complete trajectory, and 0 indicates an interrupted trajectory; the set of interrupted trajectories I is the set of all interrupted trajectory ids.

4. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 3, characterized in that: Step 3.1 The basic data of the interrupted trajectory includes vehicle position, interruption time, lane where the trajectory is interrupted, approach lane information, and the distance from the position where the interrupted trajectory is interrupted to the stop line; Wherein: f t , f lane , f id , f grid are respectively the vehicle coordinates, speed, acceleration at the interruption moment, the interruption moment, the entrance lane, the lane where the vehicle is located, the vehicle number, and the grid index where the vehicle is located; The distance from the stop line when the trajectory is interrupted can be represented by f L which is shown as follows: f L = L Entry of the approach road d The coordinates of the stop line are x p1,d , y p1,d , x p2,d , y p2,d , and the stop line can be linearly represented by a straight-line equation: b = y p1,d -k × x p1,d Where: k and b are the slope and intercept of the stop line straight-line equation respectively; l is the length of the interrupted trajectory end point coordinate from the stop line; L is the vector distance with direction. If the interrupted trajectory end point coordinate does not pass through the stop line, L is negative, otherwise L is positive.

5. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 4, characterized in that: Step 3.3 In all trajectory data Α, finding the trajectory data of the preceding vehicle needs to meet the following processes: First, judge whether there is a vehicle in the current lane: Where: vehicle = 0 indicates that there is no vehicle in the current lane; vehicle = 1 indicates that there is a vehicle in the current lane, but it does not yet indicate whether it is the preceding vehicle; According to the calculation process of Step 3.1, the distance l between the trajectory data of other vehicles on the lane and the stop line can be extracted L ; Set the initial value diff = 100: diff dist = l L - f L where: diff dist is the distance difference between the interruption trajectory at time f t and the distances of other vehicles on this lane; if the condition is met, assign diff dist to diff, and finally diff is the distance difference between the interruption trajectory at time f t and the distance to the vehicle in front; leading = 0 means there is no vehicle in front of the interruption trajectory; leading = 1 means there is a vehicle in front of the interruption trajectory; According to the above steps, the trajectory data of the leading vehicle at the interruption moment f t can be extracted.

6. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 5, characterized in that: Step 3.4 Based on all the trajectory data Α, the parameters in the IDM model can be determined: the desired speed v s , the safe headway time T, the maximum acceleration a max , the comfortable deceleration b, the acceleration exponent δ, the minimum spacing s0, the time interval Δt; According to the end coordinates and speed of the interrupted trajectory, the desired speeds v x,s , v y,s , the maximum acceleration a x,max , a y,max , the comfortable deceleration b x , b y and the minimum spacing s x , s y ; Specifically, the calculation process of the IDM model is as follows: Where: dvx is the velocity difference in the x-axis direction between the interrupted trajectory and the vehicle ahead; are the velocities in the x-axis direction of the interrupted trajectory and the vehicle ahead at t time f respectively; dx is the absolute value of the interval in the x-axis direction between the interrupted trajectory and the vehicle ahead; are the positions in the x-axis direction of the interrupted trajectory and the vehicle ahead respectively; L vehicle is the vehicle length; gap x is the desired spacing in the x-axis direction in the current state; are the acceleration, velocity, and position in the x-axis direction of the interrupted trajectory at the next moment; Δt is the time interval; Similarly, it can be calculated that the combined velocity at the next moment of the interruption trajectory can be calculated therefrom and the acceleration f t = f t + 1 Other information f id , f lane , Can be obtained according to Step 1.3 and 1.

4.

7. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 6, characterized in that: The parameters of the variable acceleration motion model are the maximum acceleration a max , the velocity response coefficient σ, the time interval Δt. According to the maximum acceleration a max and the velocity at the trajectory interruption moment f t The maximum accelerations a along the x and y axes can be determined by solving a system of equations x,max , a y,max . The calculation process is as follows; There are two solutions to this system of equations. Select the solution with the same sign as the corresponding partial velocity as the partial acceleration a x,max , a y,max ; The specific process of the variable acceleration motion model is as follows: Where: Δx is the driving distance; Similarly, it is consistent with the above calculation process; f t = f t + 1 Other information f id , f lane , Can be obtained according to Step 1.3 and 1.

4.

8. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 7, characterized in that: Step 3.9 Judge whether the vehicle is in an idle condition: Among them: mode represents the operating condition state of the vehicle ahead; mode = -1 indicates that the vehicle ahead is in a non-idle operating condition; mode = 1 indicates that the vehicle ahead is in an idle operating condition; represents the acceleration of the vehicle ahead at f t moment, with the unit of m / s; represents the speed of the vehicle ahead at f t moment, with the unit of km / h.

9. The method for repairing vehicle trajectories at signalized intersections based on roadside lidar according to claim 8, characterized in that: The formula for the distance between two vehicles in Step 3.10 is: d f,l = l L - f L where: d f,l is the distance between the interrupted trajectory and the vehicle ahead; l L is the distance from the vehicle ahead to the stop line, and the calculation process is the same as that in Step 3.2; dist = d f,l -L vehicle -s safe The acceleration of the interrupted trajectory at the next moment is: Wherein: is the speed at the interruption time f of the interrupted trajectory t ; L vehicle is the vehicle length; s safe is the minimum safety distance; the acceleration at the next moment of the interrupted trajectory; According to the trajectory interruption time f t the velocities along the x and y axes and the corresponding acceleration can be calculated It can be obtained according to the above calculation principle.

Images