Signal intersection vehicle trajectory reconstruction method based on multi-source data fusion

Abstract

The invention discloses a multi-source data fusion signalized intersection vehicle trajectory reconstruction method in an intelligent network connection environment, which comprises the following steps: acquiring motion data, signal phase data and speed camera data of an intelligent network connection vehicle (CAV) and surrounding vehicles sensed by the intelligent network connection vehicle, and determining a signal period when the vehicle passes through an intersection according to the data; the vehicle with the same period as the vehicle with the known track passes through the intersection, and the track of a common vehicle (RV) in front of the vehicle can be reconstructed in sequence from back to front by applying a track reconstruction optimization model based on an intelligent driver model (IDM); when vehicles with known trajectories passing through the intersection in the same period as the RV exist in front of the RV, vehicle trajectories are sequentially reconstructed from front to back by applying a trajectory reconstruction method based on IDM and grid search; when there is no known track vehicle passing through the intersection in the same period as the RV in front of the RV, constructing a virtual vehicle to replace the CAV based on the signal phase data; and fusing the trajectory reconstruction results of the same vehicle by using a historical weight-based trajectory fusion algorithm.

Description

Technical Field

[0001] The present invention relates to the field of intelligent traffic management, and in particular to a vehicle trajectory reconstruction method for a signalized intersection using multi-source data fusion in an intelligent network environment. Background Art

[0002] Vehicle trajectories contain rich, mineable spatiotemporal traffic information, which can be used in various traffic studies and applications, such as verification and calibration of traffic flow models, estimation of traffic congestion status, and estimation of traffic performance (travel time, delay, queue length). Therefore, obtaining vehicle trajectories has important theoretical and practical significance.

[0003] Vehicle trajectory data acquisition is generally based on fixed sensors (magnetic induction coils, video cameras, laser radars, millimeter wave radars, etc.), but fixed detectors have high installation and maintenance costs and can only collect vehicle trajectory data within a limited detection range. With the rapid development of new-generation information technologies such as artificial intelligence, mobile Internet, and big data, a new generation of intelligent transportation systems with intelligent networked vehicles as the main feature will gradually become a breakthrough in solving traffic problems. As mobile sensors distributed on the road, intelligent networked vehicles are equipped with various environmental perception sensors that can dynamically detect the movement of vehicles around them, providing a new channel for collecting vehicle trajectory data. In addition, the integration of perception data from roadside speed cameras at signalized intersections is conducive to improving the accuracy of vehicle trajectory data reconstruction. Summary of the invention

[0004] The purpose of the present invention is to provide a vehicle trajectory reconstruction method for a signalized intersection based on multi-source data fusion, and the technical problem to be solved is to address the deficiencies of the above-mentioned prior art. The trajectory reconstruction method can effectively reproduce the microscopic traffic flow trajectory within the range of a single through lane entrance at a signalized intersection with any traffic flow in an intelligent network environment.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A vehicle trajectory reconstruction method for a signalized intersection based on multi-source data fusion comprises the following steps:

[0007] Step 1: Arrange the vehicle passing data and divide the problem into two scenarios based on whether two consecutive CAVs pass through the intersection in the same signal cycle:

[0008] Step 1-1, determine the cycle corresponding to the vehicle passing through the signalized intersection: If the vehicle passes the stop line during the green light time and yellow light time of a certain signal cycle, its cycle number can be confirmed. This process can be obtained based on the detection of the stop line by the speed camera. Repeat the above process to obtain the cycle number of continuous vehicles passing through the intersection in a single straight lane. The calculation formula is as follows:

[0009]

[0010] Step 1-2: Compare the cycle numbers of the two CAVs passing through the intersection. If they are the same, go to step 2; if they are different, go to step 3.

[0011] Step 2: Reconstruct the trajectory of the RV passing through the intersection with the LRV in scenario 1:

[0012] Step 2-1: Confirm the number of vehicles to be reconstructed m i and the time domain of the trajectory to be reconstructed

[0013] Step 2-2: Set the initial position of the vehicle and set the search grid for IDM parameters and initial speed:

[0014]

[0015] Step 2-3: Use the kinematic formula to update the vehicle position, velocity and acceleration, and store the velocity through sections s1 and s2 and the position at the corresponding time into the corresponding set:

[0016]

[0017] s i，j (t) = x i，j-1 (t)-x i，j (t),

[0018] x i，j (t+Δt)=x i，j (t)+v i，j (t+Δt)Δt,

[0019] v i，j (t+Δt)=v i，j (t)+a i，j (t+Δt)Δt,

[0020]

[0021] Step 2-4: Based on the synthetic error formula, search for a set of parameters that minimizes the error as the optimal parameters, and obtain the corresponding trajectory as the optimal trajectory:

[0022]

[0023] Step 2-5: Use the newly reconstructed trajectory as the known trajectory, repeat steps 2-1 to 2-4 to continue reconstructing the subsequent vehicle trajectory, and finally select the corresponding time domain data as the trajectory data to be fused.

[0024] Step 3: Reconstruct the RV trajectory under scenario 2:

[0025] Step 3-1: Determine the number of RVs that pass through the intersection with LRV and TRV, expressed as m i1 ,m i2 .

[0026] Step 3-2: Use the method described in step 2 to reconstruct the trajectory of the RV passing through the intersection together with the LRV. The corresponding time domain is

[0027] Step 3-3: Set up a virtual car based on the signal phase data and use the virtual car as a known trajectory:

[0028]

[0029] Step 3-4: Based on the virtual vehicle trajectory, use the method described in step 2 to reconstruct the trajectory of the RV that passes through the intersection together with the TRV.

[0030] Step 3-5: Filter out the corresponding time domain The data is used as the trajectory data to be fused.

[0031] Step 4: Use the IDM-based optimization model to reconstruct the vehicle trajectory starting from TRV.

[0032] Step 4-1: Determine the time domain to be reconstructed

[0033] Step 4-2: Solve the following optimization model to reconstruct the trajectory of the preceding vehicle.

[0034]

[0035] x i，j-1 (t+Δt)=x i，j-1 (t)+v i，j-1 (t+Δt)Δt,

[0036] v i，j-1 (t) = v i，j-1 (t-Δt)+a i，j-1 (t)Δt,

[0037] b≤a i，j-1 (t)≤A,

[0038]

[0039] x i，j-1 (t)-x i，j (t)≥l,

[0040]

[0041] Step 4-3: Use Savitzky-Golay filter to smooth the vehicle trajectory to obtain the trajectory to be fused.

[0042] Step 4-4: Taking the motion state of the vehicle as known, repeat steps 4-1 to 4-3 to reconstruct the trajectory of the preceding vehicle in sequence.

[0043] Step 4-5: Filter out the data in the corresponding time domain as the trajectory data to be fused.

[0044] Step 5: Fusion the trajectories from step 2 to step 4 based on the historical weights obtained from the historical errors:

[0045] RV passing through the intersection in the same cycle as LRV:

[0046] RVs that pass through intersections at different cycles than LRVs:

[0047] The beneficial effects of the present invention are as follows: the present invention fully utilizes the communication advantages of roadside perception equipment and CAV under intelligent networking, realizes multi-source perception data sharing, and reconstructs the microscopic trajectory of continuous vehicles in a single straight lane of a signalized intersection through the trajectory reconstruction optimization model and method extended by IDM. The public data set SUMO simulation data verifies that the present invention can reconstruct high-precision vehicle trajectories under any traffic flow and CAV penetration rate, which is an effective exploration of traffic perception methods under intelligent networking environments and provides data support for future traffic management. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the process of reconstructing the RV trajectory between two consecutive CAVs according to the present invention.

[0049] Figure 2 It is a schematic diagram of an implementation scenario of the present invention.

[0050] Figure 3 Schematic diagrams of two reconstruction scenarios of the present invention.

[0051] Figure 4 It is a schematic diagram of the fusion of the reconstruction results of the present invention (taking the methods described in S2 and S4 as an example).

[0052] Figure 5 It is a schematic diagram of the trajectory reconstruction result in the example of the present invention.

[0053] Figure 6 It is a trajectory error diagram under different intelligent connected vehicle penetration rates and traffic flow conditions in the examples of the present invention. DETAILED DESCRIPTION

[0054] In order to more clearly demonstrate the purpose, technical solution and effect of the vehicle trajectory reconstruction in the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is worth noting that the scenarios described below are only some representative embodiments, not all embodiments.

[0055] like Figure 2 As shown, the present invention divides the RV trajectory between two consecutive CAVs in a single lane of a signalized intersection into two situations, namely, two CAVs pass through the intersection in the same cycle and two CAVs pass through the intersection in different cycles.

[0056] (1) Scenario 1: Two CAVs pass through an intersection in the same cycle

[0057] like Figure 3 As shown in the figure, the vehicle to be reconstructed is between the two CAVs and passes through the intersection at the same cycle as the two CAVs. Based on the LRV motion state data, the vehicle trajectory can be reconstructed from front to back using the method described in step (2). At the same time, based on the TRV motion state data, the trajectory can be reconstructed from back to front using the method described in step (4). Finally, the trajectory reconstruction results of the same RV are weighted averaged based on the historical weights to reconstruct the continuous vehicle trajectory under scenario 1.

[0058] (2) Scenario 2: Two CAVs pass through an intersection at different times

[0059] like Figure 3 As shown in the figure, the vehicle to be reconstructed is between the two CAVs, but some RVs pass through the intersection with the first CAV, while some vehicles need to pass through the intersection in subsequent cycles. Vehicles that pass through the intersection in the same cycle as the CAV can be reconstructed using the method described in scenario 1. For vehicles that pass through the intersection in subsequent cycles, a virtual vehicle can be set to replace the LRV based on the signal phase data, and the method described in step (2) can be used to reconstruct them from front to back. Finally, the trajectory reconstruction results of the same RV are weighted averaged based on the historical weights to reconstruct the continuous vehicle trajectory under scenario 2.

[0060] Before implementing this strategy, in order to simplify the problem and simulate the application of this method more realistically, the following assumptions are made:

[0061] 1) Vehicles cannot change lanes within a certain distance at the intersection entrance, so only the trajectory reconstruction problem of a single lane is considered.

[0062] 2) Based on the perception of CAVs, the motion states of the TRV and LRV of the front and rear vehicles are known, and the perception error is not considered.

[0063] 3) The speed detection accuracy of the camera is inversely proportional to the distance. Therefore, the camera is set to identify all vehicles passing through s1, s2, and s3, and to identify the speed of vehicles passing through s1 and s2.

[0064] Example 1

[0065] See also Figure 1 ,A vehicle trajectory reconstruction method for signalized intersections based on multi-source data fusion,

[0066] S1, based on the intelligent connected vehicle perception data, speed camera data, and signal phase data, determine the cycle number, sequence, time and speed of the vehicle passing through the signal intersection. Select two CAVs in sequence to determine whether they pass through the intersection in the same signal cycle, and divide the problem into two scenarios.

[0067] S2, for scenario 1, based on the motion state data of the LRV, the trajectory of the RV passing through the intersection with the LRV, specifically:

[0068] S2-1, confirm the number of vehicles to be reconstructed m i and the time domain of the trajectory to be reconstructed

[0069] S2-2, for the subsequent RVs of the LRV, set the vehicle initial position, and set the search grid for the IDM parameters and initial speed:

[0070]

[0071]

[0072] S2-3, for the first vehicle after LRV, use the kinematic formula to gradually update the vehicle from Start to The ending position, velocity and acceleration, and the velocity through sections s1 and s2 and the position at the corresponding time are stored in the corresponding collection:

[0073]

[0074] s i，j (t) = x i，j-1 (t)-x i，j (t),

[0075] x i，j (t+Δt)=x i，j (t)+v i，j (t+Δt)Δt,

[0076] v i，j (t+Δt)=v i，j (t)+a i，j (t+Δt)Δt,

[0077]

[0078] S2-4. Based on the synthetic error formula, search for a set of parameters that minimize the error as the optimal parameters, and obtain the corresponding trajectory as the optimal trajectory:

[0079]

[0080] S2-5, take the newly reconstructed trajectory of the first vehicle after the LRV as the known trajectory, repeat S2-3 to 2-4, continue to reconstruct the trajectory of subsequent vehicles, and finally The trajectory data of each vehicle at the entrance of the intersection is filtered out as the trajectory data to be fused.

[0081] S3, for scenario 2, the steps of RVs trajectory passing through the intersection with LRV are the same as S2, and the trajectory of RVs passing through the intersection with TRV needs to be reconstructed based on the signal phase setting virtual car. In addition, the time domain of the trajectory to be reconstructed is different from S2, specifically:

[0082] S3-1, determine the number of RVs that pass through the intersection together with LRV and TRV, expressed as m i1 ,m i2 .

[0083] S3-2, use the method described in S2 to reconstruct the trajectory of the RV passing through the intersection with the LRV, but the time domain needs to be modified to

[0084] S3-3, set up a virtual car based on the signal phase data, and use the virtual car as a known trajectory:

[0085]

[0086]

[0087] S3-4, based on the virtual vehicle trajectory, use the method described in step 2 to reconstruct the RV trajectory that passes through the intersection with the TRV, but the time domain needs to be modified to

[0088] S3-5, filter out the data in the corresponding time domain as the trajectory data to be fused.

[0089] S4, using the IDM-based optimization model to reconstruct the vehicle trajectory from TRV forward in sequence, specifically:

[0090] S4-1, determine the time domain to be reconstructed Note that for scenario 2, m i1 +m i2 =m i .

[0091] S4-2, input the following optimization model into the solver and solve it to reconstruct the trajectory of the preceding vehicle.

[0092]

[0093] x i，j-1 (t+Δt)=x i，j-1 (t)+v i，j-1 (t+Δt)Δt,

[0094] v i，j-1 (t) = v i，j-1 (t-Δt)+a i，j-1 (t)Δt,

[0095] b≤a i，j-1 (t)≤A,

[0096]

[0097] S4-3, use Savitzky-Golay filter to smooth the vehicle trajectory to obtain the trajectory to be fused.

[0098] S4-4, taking the motion state of the vehicle as known, repeat S4-1 to S4-3 to reconstruct the trajectory of the preceding vehicle in sequence.

[0099] S4-5, filter out the time domain corresponding to each vehicle The data is used as the trajectory data to be fused.

[0100] S5, such as Figure 4 As shown, the trajectories obtained through S2 to S4 are fused based on the historical weights obtained based on the historical errors:

[0101] RV passing through the intersection in the same cycle as LRV:

[0102] RVs that pass through intersections at different cycles than LRVs:

[0103] Example 2

[0104] In this embodiment, the vehicle trajectory reconstruction method for a signal intersection with multi-source data fusion provided in Example 1 is used to verify the microscopic vehicle trajectory data under different traffic flow conditions simulated by SUMO. The parameter values ​​required by the method are shown in Table 1:

[0105] Table 1

[0106]

[0107]

[0108] Table 2 shows the vehicle trajectory reconstruction errors when the traffic flow is 1000 vehicles / hour and the CAV penetration rate is 5%, 10%, 15%, and 20%, respectively. The errors are the average root mean square errors of the vehicle trajectories. It can be found that the reconstruction error of the present invention is small, which proves the effectiveness of the present invention. In addition, the trajectory reconstruction capability of the present invention in low saturation, under saturation, saturation, and over saturation inlet roads is tested. Figure 5-Figure 6 , which intuitively reflects its good accuracy and excellent robustness under different traffic flow conditions.

[0109] Table 2

[0110] Permeability Position error (m) Queue position error (meters) Moving position error (meter) Time error (seconds) 5％ 0.4727 0.0039 0.5346 0.0809 10％ 0.3973 0.0070 0.4485 0.0833 15％ 0.3215 0.0049 0.3604 0.0669 20％ 0.2749 0.0095 0.3018 0.0513

[0111] The above are preferred embodiments of the present invention and are not intended to limit the specific details of the present invention. For those skilled in the art, the present invention may be modified. Any modification, equivalent replacement, improvement, etc. made within the scope of the present invention shall fall within the protection scope of the present invention. The protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A vehicle trajectory reconstruction method for a signalized intersection based on multi-source data fusion, characterized in that: The following steps are involved: (1) Obtain the motion states of two adjacent intelligent connected vehicles (CAVs) and their adjacent ordinary vehicles (RVs), and determine the corresponding period of the vehicles passing through the signalized intersection based on the detection data of the speed camera and the signal phase data. The problem is divided into two scenarios based on whether the RVs with known trajectories pass through the intersection in the same period. The RVs are named LRV and TRV according to their front and rear positions. (2) Scenario 1: The LRV and TRV pass through the intersection in the same cycle. At this time, the vehicle trajectory is reconstructed from the front to the back based on the motion state of the LRV until the TRV. The details are as follows: the time when the vehicle passes through the entrance lane detected by the speed camera is used as the initial state. When the motion state of the preceding vehicle is known, the motion state of the current vehicle is continuously updated using the intelligent driver model IDM until the vehicle passes through the intersection. The IDM parameter grid is set to search for a set of IDM parameters that minimizes the synthetic error of the vehicle's motion state when passing through each detection section, and the corresponding vehicle trajectory is used as the reconstructed trajectory. The reconstructed trajectory is then used as the known data and the above process is repeated. (3) Scenario 2: LRV and TRV pass through the intersection in different cycles; at this time, multiple ordinary vehicles RVs passing through the intersection together with LRV are reconstructed based on step (2); to reconstruct the remaining RVs, a virtual vehicle is set to replace LRV based on the signal phase data; the first vehicle passing through the intersection in the same cycle as TRV decelerates because it cannot pass through the intersection during the green light time of the previous cycle until it stops in front of the stop line. During this period, it is assumed that there is a virtual vehicle stopped in front of the stop line; when the green light comes on in the next cycle, the vehicle starts to accelerate. At this time, the acceleration behavior of the vehicle is equivalent to following a virtual vehicle at an infinite distance; by taking the trajectory of the virtual vehicle as the known trajectory, the subsequent vehicle trajectory can be reconstructed using step (2); (4) Using the IDM-based optimization model, we sequentially reconstruct the trajectories of RVs that pass through the intersection in the same cycle as the TRV, starting from the TRV. (5) Based on the historical error data statistics, historical weight data is obtained, and the trajectories of the same vehicle reconstructed in steps (2) and (4), and in steps (3) and (4) are weighted averaged to obtain the final vehicle trajectory; (6) Repeating steps (1) to (5) can reconstruct the continuous traffic flow micro-trajectory at the entrance of the signalized intersection.

2. The signalized intersection vehicle trajectory reconstruction method based on multi-source data fusion according to claim 1 is characterized in that: Multi-source data includes speed camera perception data, CAV perception data, and signal phase data; among them, the speed camera perception data is the two speed cameras located above the entrance road, which detect the speed and time of vehicles passing through the starting section of the entrance road, the middle section of the solid line, and the section where the stop line is located; the CAV perception data is the motion state data of the intelligent connected vehicle and its perception vehicle, including the position, speed, and acceleration in continuous time; the signal phase data is the phase information of each signal at the corresponding signal intersection.

3. The signalized intersection vehicle trajectory reconstruction method based on multi-source data fusion according to claim 1 is characterized in that: The reconstructed scene is a single through lane in a signalized intersection, and the reconstructed object is the spatiotemporal trajectory of vehicles in the scene that are not perceived by the CAV.

4. The signalized intersection vehicle trajectory reconstruction method based on multi-source data fusion according to claim 1 is characterized in that: The cycle number G of the vehicle passing through the signalized intersection in step (1) is i,j Calculation method: in, is the vehicle V i,j The time to pass the intersection stop line s1; C is the signal cycle of the signal intersection; C i,j is the vehicle V i,j The cycle number through the intersection.

5. The signalized intersection vehicle trajectory reconstruction method based on multi-source data fusion according to claim 1 is characterized in that: In step (2), the vehicle in scene 1 is moved from the front vehicle V i,j-1 Reconstruct the rear car V i,j The specific calculation formula is: s i,j (t)=x i,j-1 (t)-x i,j (t), x i,j (t+Δt)=x i,j (t)+v i,j (t+Δt)Δt, v i,j (t+Δt)=v i,j (t)+a i,j (t+Δt)Δt, Among them, s k , k = 1, 2, 3, respectively representing the three sections detected by the speed camera, namely the section where the stop line is located, the section where the solid line is located, and the section where the entrance road starts; For section s k The location coordinates of x i,j (t),v i,j (t),a i,j (t),s i,j (t), Respectively represent V i,j Position coordinates, velocity, acceleration, headway, and expected headway at time t; Indicates V i,j Through the section k time; A,b,v f ,δ,l,s0,T0 are IDM parameters, which represent the maximum acceleration, minimum speed bump, free flow speed, acceleration index, vehicle length, minimum headway, and minimum headway time, respectively; X(i,j,μ) represents V i,j IDM parameters used; lb(X), ub(X), n(X) represent the lower limit, upper limit and number of search grids of X respectively; A * ,b * , Represents the optimal parameters found by grid search; Δt represents the time step; and Used to store the corresponding and m i Indicates the number of RVs between CAVs. Vehicles are numbered starting from the front CAV with subscript j=0 and ending at the rear CAV with subscript j=m i +1; from Start counting to That is, V i,j The trajectory of V i,j The trajectory of the rear vehicle V continues to be reconstructed i,j+1 The above process is repeated until all RV trajectories are reconstructed and finally the The trajectory data is used to reproduce the space-time trajectory of the vehicle at the entrance.

6. The signalized intersection vehicle trajectory reconstruction method based on multi-source data fusion according to claim 1 is characterized in that: The motion state of the virtual car in step (3) is expressed as: Among them, m i1 ,m i2 Respectively represent the number of RVs passing through the intersection at the same phase as TRV and LRV, m i1 +m i2 =m i ; v i,vir (t),x i,vir (t) represent the speed and position coordinates of the virtual car respectively; t i+1.g Indicates the green light start time of the cycle in which the rear CAV passes through the signalized intersection; It indicates the time when the first vehicle in the queue passes through section s3; It indicates the time when the previous RV of TRV passes through section s1; f t Indicates the correction time for the vehicle to start accelerating before the green light comes on; f d is the correction value of the virtual car position.

7. The method for reconstructing vehicle trajectories at a signalized intersection based on multi-source data fusion according to claim 1, characterized in that: The objective function of the IDM-based optimization model in step (4) is: Among them, e i,j-1 (t) represents the reconstruction error; The constraints of the IDM-based optimization model include IDM dynamic constraints, outlier correction constraints, general dynamic constraints, vehicle spacing constraints, detection section speed and position constraints, initial headway constraints, and initial position constraints; The IDM dynamic constraints are: in, The outlier correction constraint is: Among them, th is used to determine whether correction is needed. The threshold value of The general dynamic constraints are: x i,j-1 (t+Δt)=x i,j-1 (t)+v i,j-1 (t+Δt)Δt, v i,j-1 (t)=v i,j-1 (t-Δt)+a i,j-1 (t)Δt, b≤a i,j-1 (t)≤A, in, Respectively represent the lower and upper limits of the acceleration change rate; The vehicle spacing constraint is: x i,j-1 (t)-x i,j (t)≥l; The speed and position constraints of the detection section are: Among them, δ v It represents the error of speed detection by speed camera. Speed ​​camera can obtain the speed of vehicle passing through section s k time, and thus the position of the vehicle at the corresponding moment can be determined. In addition, since speed detection is negatively correlated with distance, the camera only detects the speed of the vehicle passing through sections s1 and s2; The initial headway constraint is: The initial position constraints are: The vehicle motion trajectory can be solved according to the objective function and constraints. Then, the vehicle position data is smoothed using the Savitzky-Golay filter and the vehicle motion state is taken as known to continue reconstructing the front vehicle trajectory.

8. The signalized intersection vehicle trajectory reconstruction method based on multi-source data fusion according to claim 1 is characterized in that: The historical weight data in step (5) is obtained by calculating the error in the historical vehicle trajectory data by the method described in claims 5, 6, and 7. The error calculation method is as follows: In the method of claims 5 and 6, a vehicle V passing through the intersection together with the LRV i,j The historical error can be expressed as: in, represents the average error of j RVs away from LRV; Claim 6: The method according to claim 6: There is no V that passes through the intersection together with the LRV i,j The historical error can be expressed as: in, Indicates the distance from the virtual car to jm i1 Average error of RVs; Rights 7 in V i,j The historical error can be expressed as: in, It represents the average error of RVs j away from TRV.

9. The method for reconstructing vehicle trajectories at a signalized intersection based on multi-source data fusion according to claim 1, characterized in that: The calculation method of historical weight data and trajectory fusion in step (5) is as follows: The historical weights of all RVs in scenario 1 and some RVs passing through the intersection together with LRVs in scenario 2 are: in, They represent the weights of the trajectories reconstructed by fusing the methods described in claim 5 and claim 7 respectively; Represents a trajectory reconstructed by the method of claim 5 and claim 7; The historical weight of the RV that did not pass the intersection with the LRV in scenario 2: in, They represent the weights of the trajectory reconstructed by fusing the methods described in claim 6 and claim 7, respectively. Represents the trajectory reconstructed by the method described in claim 6 and claim 7.

Images