A vehicle blind area trajectory prediction method and device in a highway network

Abstract

This invention relates to the field of intelligent transportation technology and provides a method and apparatus for predicting vehicle blind spot trajectories in a highway network. In this embodiment, for a target area of ​​the highway, a road network topology map is first constructed using sensing devices and road segment physical parameters of the target area. For vehicles in the blind spot of this target area, entrance and exit observation data are collected to determine an initial path set. Then, the physical reachability parameters of nodes within the blind spot are determined using the road network topology map, and invalid paths in the initial path set are deleted using these parameters. Finally, the running trajectory of each vehicle in the blind spot is predicted using the deleted path set. This method allows for the reconstruction of vehicle trajectories even when a direct correspondence based on a unique vehicle identifier cannot be established.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and in particular to a method and apparatus for predicting vehicle blind spot trajectories in a highway network. Background Technology

[0002] With the continuous increase in the mileage of highways and the constant rise in traffic flow, traffic managers have an increasingly urgent need for refined perception of vehicle operating status. Vehicle trajectory data serves as the foundation for traffic situation analysis, event detection, and proactive control, and its completeness and accuracy directly affect the effectiveness of intelligent transportation systems.

[0003] During the construction of highway sensing facilities, continuous coverage across the entire road section is difficult to achieve due to limitations such as construction costs, pole resources, and power and communication conditions. Taking the western section of a certain ring expressway as an example, only two variable message signs and supporting sensing facilities are deployed along a 15-kilometer stretch from Gaoqiao to the northern section of the expressway, resulting in a sensing coverage rate of less than 15%. This low-density sensing layout leads to numerous blind spots, making it impossible to directly observe the driving trajectories of vehicles within these blind spots.

[0004] Furthermore, the sensing facilities at the entrance and exit sections of each blind spot may fail to accurately identify the unique vehicle identifier (such as the license plate number) due to equipment malfunction, environmental interference, or damaged license plates. In this case, even if the entrance and exit sections collect vehicle passage records, it is impossible to directly link the entrance and exit records of the same vehicle through the license plate, resulting in a double dilemma of "unknown origin and destination" for vehicle trajectory reconstruction. Summary of the Invention

[0005] In view of this, this application provides a method and apparatus for predicting vehicle blind spot trajectories in a highway network to solve the problem that vehicle trajectories cannot be restored when vehicle information cannot be identified.

[0006] The first aspect of this application provides a method for predicting vehicle blind spot trajectories in a highway network, applied to an intelligent transportation platform, the method comprising: Determine the physical parameters of road segments and the deployment points of sensing facilities in the target area of ​​the highway, and construct the road network topology map G = (V, E) of the target area, where V is the set of nodes determined based on the deployment points of the sensing facilities, and E is the set of road segments determined based on the nodes and the physical parameters of the road segments; For any blind spot, determine the target vehicle and entrance / exit observation data within the blind spot, and then determine the initial path set of the target vehicle based on the entrance / exit observation data; Based on the road network topology map, the physical reachability parameters of the nodes in the blind zone are determined. Based on the road network topology map and the physical reachability parameters, invalid paths are identified from the initial path set and removed to obtain the final path set. Finally, the running trajectory of each target vehicle is estimated through the final path set.

[0007] Optionally, the target vehicle is a vehicle with unidentifiable features, and the step of determining and removing invalid paths from the initial path set based on the road network topology map and the physical reachability parameters includes: For any entrance target vehicle i, determine and associate it with its corresponding exit target vehicle j, and determine the passage time of the entrance target vehicle i and the exit target vehicle j. Then, based on the physical accessibility parameters, the passage time interval of the blind zone is determined. When the passage time is not within the passage time interval, the path of the target entrance vehicle i and the target exit vehicle j is determined as an invalid path and removed. The necessary nodes for the target vehicle are determined by the physical parameters of the road segment, and invalid paths are identified and removed based on these necessary nodes.

[0008] Optionally, the target vehicle is a vehicle with unidentifiable features, and the road segment physical parameters include lane position; The step of determining and removing invalid paths from the initial path set based on the road network topology map and the target physical reachability parameters includes: For any entrance target vehicle i, determine and associate it with its corresponding exit target vehicle j. Determine the entrance lane position l_i and exit lane position l'_j of the entrance target vehicle i and the exit target vehicle j. Calculate the minimum number of lane changes based on the entrance lane position and the exit lane position. If the minimum number of lane changes is greater than the maximum number of lane changes determined based on the road network topology map, then the path of the target entrance vehicle i and the exit target vehicle j is determined as an invalid path and removed.

[0009] Optionally, determine the passage time for each pair of inbound target vehicle i and outbound target vehicle j. Subsequently, the method further includes: The slope compensation time is determined by the slope information s in the physical parameters of the road section, and then the passage time interval of the blind spot is corrected by the slope compensation time.

[0010] Optionally, the step of determining and associating any entrance target vehicle i with its corresponding exit target vehicle j includes: The fingerprint feature vector of each vehicle in the blind zone is determined. The fingerprint feature vector includes a time sequence fingerprint determined based on the entry time series and exit time series of each vehicle, a vehicle model sequence fingerprint determined based on the entry vehicle model sequence and exit vehicle model sequence, a speed correlation fingerprint determined based on the entry speed sequence and exit speed sequence, and a lane position fingerprint determined based on the entry lane position and exit lane position. Based on the fingerprint feature vector, several groups of inbound target vehicles and outbound target vehicles are identified and associated.

[0011] A second aspect of this application provides a vehicle blind spot trajectory prediction device for a highway network, applied to an intelligent transportation platform, the device comprising: The topology graph construction unit is used to determine the physical parameters of road segments and the deployment points of sensing facilities in the target area of ​​the highway, and to construct the road network topology graph G = (V, E) of the target area, where V is the set of nodes determined based on the deployment points of the sensing facilities, and E is the set of road segments determined based on the nodes and the physical parameters of the road segments. An initial path determination unit is used to determine, for any blind zone, the target vehicle and entrance / exit observation data within the blind zone, and then determine the initial path set of the target vehicle based on the entrance / exit observation data. The trajectory determination unit is used to determine the physical reachability parameters of nodes in the blind zone based on the road network topology map, determine invalid paths from the initial path set and remove them based on the road network topology map and the physical reachability parameters to obtain a final path set, and finally predict the trajectory of each target vehicle through the final path set.

[0012] Optionally, the target vehicle in the initial path determination unit and the trajectory determination unit is a vehicle with unidentifiable features; The trajectory determination unit is used to determine and associate any entrance target vehicle i with its corresponding exit target vehicle j, and to determine the passage time of the entrance target vehicle i and the exit target vehicle j. Then, based on the physical accessibility parameters, the passage time interval of the blind spot is determined. When the passage time is not within the passage time interval, the path of the target entrance vehicle i and the target exit vehicle j is determined as an invalid path and removed. The necessary nodes for the target vehicle are determined by the physical parameters of the road segment, and invalid paths are identified and removed based on these necessary nodes.

[0013] Optionally, the target vehicle in the initial path determination unit and the trajectory determination unit is a vehicle with unidentifiable features, and the road segment physical parameters include lane position; The trajectory determination unit is used to determine and associate any entrance target vehicle i with its corresponding exit target vehicle j, determine the entrance lane position l_i and the exit lane position l'_j of the entrance target vehicle i and the exit target vehicle j, and calculate the minimum number of lane changes based on the entrance lane position and the exit lane position. When the minimum number of lane changes is greater than the maximum number of lane changes determined based on the road network topology map, the path of the target entrance vehicle i and the exit target vehicle j is determined as an invalid path and removed.

[0014] Optionally, the device further includes: The time correction unit is used to determine the passage time for each pair of inbound target vehicles i and outbound target vehicles j. Then, the slope compensation time is determined by the slope information s in the physical parameters of the road section, and the passage time interval of the blind spot is corrected by the slope compensation time.

[0015] Optionally, the trajectory determination unit is used to determine the fingerprint feature vector of each vehicle in the blind zone. The fingerprint feature vector includes a time sequence fingerprint determined based on the entry time series and exit time series of each vehicle, a vehicle model sequence fingerprint determined based on the entry vehicle model sequence and exit vehicle model sequence, a speed correlation fingerprint determined based on the entry speed sequence and exit speed sequence, and a lane position fingerprint determined based on the entry lane position and exit lane position. Based on the fingerprint feature vector, several groups of inbound target vehicles and outbound target vehicles are identified and associated.

[0016] In the embodiments provided in this application, for a highway target area, a road network topology map is first constructed using sensing devices and road segment physical parameters of the target area. For vehicles in blind spots within the target area, entrance and exit observation data are collected to determine an initial path set. Then, the physical reachability parameters of nodes within the blind spot are determined using the road network topology map, and invalid paths in the initial path set are deleted using these parameters. Finally, the running trajectory of each vehicle in the blind spot is estimated using the deleted path set. This method allows for the reconstruction of vehicle trajectories even when a direct correspondence based on a unique vehicle identifier cannot be established. Attached Figure Description

[0017] Figure 1 A flowchart illustrating the method provided in this application embodiment; Figure 2 This is a structural diagram of the device provided in the embodiments of this application; Figure 3 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0019] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0020] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0021] This application provides a method and apparatus for predicting vehicle blind spot trajectories in a highway network to solve the problem that vehicle trajectories cannot be restored when vehicle information cannot be identified.

[0022] The technical solutions of this application will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0023] like Figure 1 The diagram shown is a flowchart of a vehicle blind spot trajectory prediction method in a highway network provided in this application. The process may include the following steps: Step S101: Determine the physical parameters of the road segments and the deployment points of sensing facilities in the target area of ​​the highway, and construct the road network topology map G = (V, E) of the target area, where V is the set of nodes determined based on the deployment points of the sensing facilities, and E is the set of road segments determined based on the nodes and the physical parameters of the road segments.

[0024] In this embodiment, it is first necessary to select the range of highway sections for which the road network topology needs to be constructed, such as the western section of Ningbo Ring Expressway, from Gaoqiao Interchange (K0+000) to Ningbo North Interchange (K15+200), and then use this range as the target area in this application.

[0025] Secondly, the deployment points of various sensing facilities within the area are determined. These include license plate recognition checkpoints, radar detectors, integrated radar-visual machines, ETC gantries, and variable message signs. The location coordinates of each facility are recorded, and each facility is then treated as a node in the road network topology map, thus determining the node set V. This is shown in Table 1 below. Table 1 Finally, first determine the physical parameters of the road segments within the area. These parameters include the segment length L, the number of lanes n, lane function (straight / turning / emergency), ramp type (directional / roundabout / cloverleaf), maximum and minimum speed limits, ramp height and weight limits, etc. Then, divide the road segments based on the above nodes and determine the physical parameters of each segment in conjunction with the aforementioned physical parameters. See the table below: Table 2 It should be noted that there are many methods for dividing road segments, not limited to the nodes mentioned above. For example, road segments can also be divided based on lane function change points, road geometry change points, and so on.

[0026] After determining the set of nodes V and the set of road segments E, the set of nodes V is used as the vertex set of the topology graph, and the set of road segments E is used as the edge set of the topology graph, thus forming the road network topology graph described above.

[0027] Step S102: For any blind spot, determine the target vehicle and entrance / exit observation data within the blind spot, and then determine the initial path set of the target vehicle based on the entrance / exit observation data.

[0028] In this embodiment, areas that cannot be captured by cameras along the highway can be designated as blind spots. Then, vehicles observed by cameras at the entrance and exit sections of the blind spots are designated as target vehicles, and data such as the time of entry and exit from the blind spots, lane entry and exit, speed entry and exit, and license plate number of each target vehicle are determined and used as entrance and exit observation data.

[0029] Once determined, all possible driving paths within the blind zone are determined based on the entrance and exit observation data. In other words, the initial path set contains all topologically feasible paths for each vehicle from the entrance node to the exit node.

[0030] Step S103: Determine the physical reachability parameters of the nodes in the blind zone based on the road network topology map; determine and remove invalid paths from the initial path set based on the road network topology map and the physical reachability parameters to obtain the final path set; and finally estimate the running trajectory of each target vehicle based on the final path set.

[0031] In this embodiment, the physical reachability parameter indicates whether two nodes are reachable based on existing data. That is, it uses the vehicle's actual travel time and the physical limits of the road segment to eliminate paths that are not feasible in terms of time. If the minimum travel time of a path is greater than the vehicle's actual travel time, or the maximum travel time is less than the vehicle's actual travel time, then the path is physically impossible for the vehicle to travel and should be excluded. Therefore, the physical reachability parameter can include minimum travel time and maximum travel time, which are determined comprehensively based on data such as the road segment's maximum speed limit, minimum speed limit, ramp acceleration / deceleration losses, and toll station waiting time. Paths in the initial path set whose travel time does not meet this condition are identified as invalid paths and deleted.

[0032] After deletion, if there are multiple candidate matching relationships for each target vehicle, the matching relationship with the smallest deviation between the travel time and the historical benchmark value is selected as the final trajectory.

[0033] In another embodiment, the target vehicle is a vehicle with unidentifiable features, and the step of determining and removing invalid paths from the initial path set based on the road network topology map and the physical reachability parameters includes: For any entrance target vehicle i, determine and associate it with its corresponding exit target vehicle j, and determine the passage time of the entrance target vehicle i and the exit target vehicle j. Then, based on the physical accessibility parameters, the passage time interval of the blind zone is determined. When the passage time is not within the passage time interval, the path of the target entrance vehicle i and the target exit vehicle j is determined as an invalid path and removed. The necessary nodes for the target vehicle are determined by the physical parameters of the road segment, and invalid paths are identified and removed based on these necessary nodes.

[0034] Due to issues such as damaged, obscured, or unclear license plates due to fog or haze, license plates may be unrecognizable, making it impossible to directly associate entrance and exit vehicles. Therefore, sets are constructed for these unassociated vehicles: I` = {i`1, i`2, ..., i`_m} for entrance vehicles and O` = {o`1, o`2, ..., o`_n} for exit vehicles. These are then combined with the already identified entrance and exit vehicle sets I = {i1, i2, ..., i_m} and O = {o1, o2, ..., o_n}.

[0035] Since the vehicles in sets I' and O' have no direct relationship, it is necessary to first associate the vehicles in these two sets with other sets.

[0036] Vehicles can be associated using factors such as color and model. For example, if the entrance vehicle i`2 is a white sedan, and the exit vehicles o`3, o`8, and o2 are also a white sedan, and there are no vehicles associated with o2 in set I, then three sets of entrance and exit vehicle relationships can be determined: {i`2, o`3}, {i`2, o`8}, and {i`2, o2}. After associating the vehicles using this method, the initial path set described above is constructed. Then, invalid paths are identified and deleted using the aforementioned physical reachability parameters.

[0037] Then, based on the vehicle's characteristics, the specific nodes it must pass through (such as interchanges and ETC gantries) are determined. If the vehicle cannot pass through these nodes in time, the corresponding matching pair is excluded. If there is a unique interconnection between the entrance and exit of the blind spot, the path that does not pass through this interaction is determined as an invalid path and deleted.

[0038] After deleting several invalid paths using this method, if there are multiple candidate matching relationships for each target vehicle, the matching relationship with the smallest deviation between the travel time and the historical benchmark value is selected as the final trajectory, thereby determining the blind spot driving path of each of the aforementioned target vehicles that cannot be recognized by license plate recognition.

[0039] In another embodiment, the target vehicle is a vehicle with unidentifiable features, and the road segment physical parameters include lane position; The step of determining and removing invalid paths from the initial path set based on the road network topology map and the target physical reachability parameters includes: For any entrance target vehicle i, determine and associate it with its corresponding exit target vehicle j. Determine the entrance lane position l_i and exit lane position l'_j of the entrance target vehicle i and the exit target vehicle j. Calculate the minimum number of lane changes based on the entrance lane position and the exit lane position. If the minimum number of lane changes is greater than the maximum number of lane changes determined based on the road network topology map, then the path of the target entrance vehicle i and the exit target vehicle j is determined as an invalid path and removed.

[0040] Besides travel time constraints, lane position is another important physical constraint. Therefore, this embodiment determines invalid paths by using the maximum and minimum number of lane changes. Lane change constraints in a dense solid line scenario are shown in the table below: Table 3 The lane observation data is as follows: Table 4 Using the above method, all matching pairs from entrance lane 1 to exit lane 4 are eliminated. Furthermore, if lane function division exists within the blind spot, such as lane 4 of road segment e02 being an exit ramp, and if the entrance lane is 3 and a vehicle needs to exit, a lane change is mandatory.

[0041] This embodiment uses lane location as a matching constraint for vehicles whose features cannot be identified, increasing the physical basis for matching and thus further deleting invalid paths.

[0042] In another embodiment, the passage time for each pair of inbound target vehicle i and outbound target vehicle j is determined. Subsequently, the method further includes: The slope compensation time is determined by the slope information s in the physical parameters of the road section, and then the passage time interval of the blind spot is corrected by the slope compensation time.

[0043] In this embodiment, since some road sections have uphill sections and the actual speed on the uphill sections is much lower than the speed limit, this embodiment can compensate for the minimum travel time according to the slope, thereby improving the accuracy of path recognition.

[0044] In another embodiment, determining and associating any entrance target vehicle i with its corresponding exit target vehicle j includes: The fingerprint feature vector of each vehicle in the blind zone is determined. The fingerprint feature vector includes a time sequence fingerprint determined based on the entry time series and exit time series of each vehicle, a vehicle model sequence fingerprint determined based on the entry vehicle model sequence and exit vehicle model sequence, a speed correlation fingerprint determined based on the entry speed sequence and exit speed sequence, and a lane position fingerprint determined based on the entry lane position and exit lane position. Based on the fingerprint feature vector, several groups of inbound target vehicles and outbound target vehicles are identified and associated.

[0045] This embodiment provides an association method other than association based on vehicle features. The specific implementation process is as follows: First, the time correlation is determined by the correspondence between the entry and exit time series of each vehicle. Under normal traffic conditions, the entry and exit time series maintain the same order and spacing.

[0046] Second, based on the matching degree between the inbound and outbound vehicle vehicle sequences, the ratio of passenger cars and trucks should be basically consistent.

[0047] Third, based on the correlation between the entrance speed distribution and the exit speed distribution, there is a correlation between the entry speed and the exit speed of the same vehicle.

[0048] Fourth, the relationship between the distribution of entrance lanes and exit lanes: there is a constraint relationship between the lanes a vehicle enters and the lanes it exits.

[0049] Based on the aforementioned association conditions, unassociated entry and exit vehicles are probabilistically associated from vehicle sets I', O', I, and O. Compared to direct association based on vehicle features, this method achieves higher accuracy and generates less associated data, thereby reducing the number of paths in the initial path set. This reduces computational load while improving accuracy.

[0050] This concludes the process. Figure 1 The process is shown below.

[0051] In this embodiment, for a highway target area, a road network topology map is first constructed using sensing devices and road segment physical parameters of the target area. For vehicles in blind spots within the target area, entrance and exit observation data are collected to determine an initial path set. Then, the physical reachability parameters of nodes within the blind spot are determined using the road network topology map, and invalid paths in the initial path set are deleted using these parameters. Finally, the trajectory of each vehicle in the blind spot is estimated using the deleted path set. This method allows for the reconstruction of vehicle trajectories even when a direct correspondence based on a unique vehicle identifier cannot be established.

[0052] This application also provides a vehicle blind spot trajectory prediction device for highway networks, such as...Figure 2 As shown, the device includes: The topology construction unit 201 is used to determine the physical parameters of road segments and the deployment points of sensing facilities in the target area of ​​the highway, and construct the road network topology G = (V, E) of the target area, where V is the set of nodes determined according to the deployment points of the sensing facilities, and E is the set of road segments determined according to the nodes and the physical parameters of the road segments. The initial path determination unit 202 is used to determine the target vehicle and entrance / exit observation data within any blind zone, and then determine the initial path set of the target vehicle based on the entrance / exit observation data. The trajectory determination unit 203 is used to determine the physical reachability parameters of nodes in the blind zone according to the road network topology map, determine invalid paths from the initial path set and remove them according to the road network topology map and the physical reachability parameters to obtain a final path set, and finally predict the trajectory of each target vehicle through the final path set.

[0053] In another embodiment, the target vehicle in the initial path determination unit and the trajectory determination unit is a vehicle with unidentifiable features; The trajectory determination unit is used to determine and associate any entrance target vehicle i with its corresponding exit target vehicle j, and to determine the passage time of the entrance target vehicle i and the exit target vehicle j. Then, based on the physical accessibility parameters, the passage time interval of the blind spot is determined. When the passage time is not within the passage time interval, the path of the target entrance vehicle i and the target exit vehicle j is determined as an invalid path and removed. The necessary nodes for the target vehicle are determined by the physical parameters of the road segment, and invalid paths are identified and removed based on these necessary nodes.

[0054] In another embodiment, the target vehicle in the initial path determination unit and the trajectory determination unit is a vehicle with unidentifiable features, and the road segment physical parameters include lane position; The trajectory determination unit is used to determine and associate any entrance target vehicle i with its corresponding exit target vehicle j, determine the entrance lane position l_i and the exit lane position l'_j of the entrance target vehicle i and the exit target vehicle j, and calculate the minimum number of lane changes based on the entrance lane position and the exit lane position. When the minimum number of lane changes is greater than the maximum number of lane changes determined based on the road network topology map, the path of the target entrance vehicle i and the exit target vehicle j is determined as an invalid path and removed.

[0055] In another embodiment, the device further includes: Time correction unit 204 is used to determine the passage time of each pair of entrance target vehicles i and exit target vehicles j. Then, the slope compensation time is determined by the slope information s in the physical parameters of the road section, and the passage time interval of the blind spot is corrected by the slope compensation time.

[0056] In another embodiment, the trajectory determination unit is used to determine the fingerprint feature vector of each vehicle in the blind zone. The fingerprint feature vector includes a time sequence fingerprint determined based on the entry time sequence and exit time sequence of each vehicle, a vehicle model sequence fingerprint determined based on the entry vehicle model sequence and exit vehicle model sequence, a speed correlation fingerprint determined based on the entry speed sequence and exit speed sequence, and a lane position fingerprint determined based on the entry lane position and exit lane position. Based on the fingerprint feature vector, several groups of inbound target vehicles and outbound target vehicles are identified and associated.

[0057] The present invention provides a method for predicting vehicle blind spot trajectories in a highway network in the above embodiments, and provides a device for predicting vehicle blind spot trajectories in a highway network based on the method. The above method and device can solve the problem that vehicle trajectories cannot be restored when vehicle information cannot be identified.

[0058] This embodiment also discloses a computer device, such as... Figure 3 As shown, the computer device includes a processor and a memory, the memory storing at least one instruction, which is loaded and executed by the processor to implement the vehicle blind spot trajectory prediction method in any of the above-described highway networks.

[0059] Furthermore, in the above-described embodiment of the vehicle blind spot trajectory prediction device in the highway network, the logical division of each program module is merely illustrative. In practical applications, the above functions can be assigned to different program modules as needed, for example, for the sake of corresponding hardware configuration requirements or the convenience of software implementation. That is, the internal structure of the vehicle blind spot trajectory prediction device in the highway network can be divided into different program modules to complete all or part of the functions described above.

[0060] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for predicting vehicle blind spot trajectories in a highway network, characterized in that, Applied to intelligent transportation platforms, the method includes: Determine the physical parameters of road segments and the deployment points of sensing facilities in the target area of ​​the highway, and construct the road network topology map G = (V, E) of the target area, where V is the set of nodes determined based on the deployment points of the sensing facilities, and E is the set of road segments determined based on the nodes and the physical parameters of the road segments; For any blind spot, determine the target vehicle and entrance / exit observation data within the blind spot, and then determine the initial path set of the target vehicle based on the entrance / exit observation data; Based on the road network topology map, the physical reachability parameters of the nodes in the blind zone are determined. Based on the road network topology map and the physical reachability parameters, invalid paths are identified from the initial path set and removed to obtain the final path set. Finally, the running trajectory of each target vehicle is estimated through the final path set.

2. The method according to claim 1, characterized in that, The target vehicle is a vehicle with unidentifiable features. The step of determining and removing invalid paths from the initial path set based on the road network topology map and the physical reachability parameters includes: For any entrance target vehicle i, determine and associate it with its corresponding exit target vehicle j, and determine the passage time of the entrance target vehicle i and the exit target vehicle j. Then, based on the physical accessibility parameters, the passage time interval of the blind spot is determined. When the passage time is not within the passage time interval, the path of the target entrance vehicle i and the target exit vehicle j is determined as an invalid path and removed. The necessary nodes for the target vehicle are determined by the physical parameters of the road segment, and invalid paths are identified and removed based on these necessary nodes.

3. The method according to claim 1, characterized in that, The target vehicle is a vehicle with unidentifiable features, and the physical parameters of the road segment include lane position; The step of determining and removing invalid paths from the initial path set based on the road network topology map and the target physical reachability parameters includes: For any entrance target vehicle i, determine and associate it with its corresponding exit target vehicle j. Determine the entrance lane position l_i and exit lane position l'_j of the entrance target vehicle i and the exit target vehicle j. Calculate the minimum number of lane changes based on the entrance lane position and the exit lane position. If the minimum number of lane changes is greater than the maximum number of lane changes determined based on the road network topology map, then the path of the target entrance vehicle i and the exit target vehicle j is determined as an invalid path and removed.

4. The method according to claim 2, characterized in that, Determine the passage time for each pair of inbound target vehicle i and outbound target vehicle j. Subsequently, the method further includes: The slope compensation time is determined by the slope information s in the physical parameters of the road section, and then the passage time interval of the blind spot is corrected by the slope compensation time.

5. The method according to claim 2 or 3, characterized in that, The step of determining and associating any entrance target vehicle i with its corresponding exit target vehicle j includes: The fingerprint feature vector of each vehicle in the blind zone is determined. The fingerprint feature vector includes a time sequence fingerprint determined based on the entry time series and exit time series of each vehicle, a vehicle model sequence fingerprint determined based on the entry vehicle model sequence and exit vehicle model sequence, a speed correlation fingerprint determined based on the entry speed sequence and exit speed sequence, and a lane position fingerprint determined based on the entry lane position and exit lane position. Based on the fingerprint feature vector, several groups of inbound target vehicles and outbound target vehicles are identified and associated.

6. A vehicle blind spot trajectory prediction device in a highway network, characterized in that, The device, applied to an intelligent transportation platform, includes: The topology graph construction unit is used to determine the physical parameters of road segments and the deployment points of sensing facilities in the target area of ​​the highway, and to construct the road network topology graph G = (V, E) of the target area, where V is the set of nodes determined based on the deployment points of the sensing facilities, and E is the set of road segments determined based on the nodes and the physical parameters of the road segments. An initial path determination unit is used to determine, for any blind zone, the target vehicle and entrance / exit observation data within the blind zone, and then determine the initial path set of the target vehicle based on the entrance / exit observation data. The trajectory determination unit is used to determine the physical reachability parameters of nodes in the blind zone based on the road network topology map, determine invalid paths from the initial path set and remove them based on the road network topology map and the physical reachability parameters to obtain a final path set, and finally predict the trajectory of each target vehicle through the final path set.

7. The apparatus according to claim 6, characterized in that, The target vehicles in the initial path determination unit and the trajectory determination unit are vehicles with unidentifiable features; The trajectory determination unit is used to determine and associate any entrance target vehicle i with its corresponding exit target vehicle j, and to determine the passage time of the entrance target vehicle i and the exit target vehicle j. Then, based on the physical accessibility parameters, the passage time interval of the blind spot is determined. When the passage time is not within the passage time interval, the path of the target entrance vehicle i and the target exit vehicle j is determined as an invalid path and removed. The necessary nodes for the target vehicle are determined by the physical parameters of the road segment, and invalid paths are identified and removed based on these necessary nodes.

8. The apparatus according to claim 6, characterized in that, The target vehicles in the initial path determination unit and the trajectory determination unit are vehicles with unidentifiable features, and the road segment physical parameters include lane positions; The trajectory determination unit is used to determine and associate any entrance target vehicle i with its corresponding exit target vehicle j, determine the entrance lane position l_i and the exit lane position l'_j of the entrance target vehicle i and the exit target vehicle j, and calculate the minimum number of lane changes based on the entrance lane position and the exit lane position. When the minimum number of lane changes is greater than the maximum number of lane changes determined based on the road network topology map, the path of the target entrance vehicle i and the exit target vehicle j is determined as an invalid path and removed.

9. The apparatus according to claim 7, characterized in that, The device further includes: The time correction unit is used to determine the passage time for each pair of inbound target vehicles i and outbound target vehicles j. Then, the slope compensation time is determined by the slope information s in the physical parameters of the road section, and the passage time interval of the blind spot is corrected by the slope compensation time.

10. The apparatus according to claim 7 or 8, characterized in that, The trajectory determination unit is used to determine the fingerprint feature vector of each vehicle in the blind zone. The fingerprint feature vector includes a time sequence fingerprint determined based on the entry time series and exit time series of each vehicle, a vehicle model sequence fingerprint determined based on the entry vehicle model sequence and exit vehicle model sequence, a speed correlation fingerprint determined based on the entry speed sequence and exit speed sequence, and a lane position fingerprint determined based on the entry lane position and exit lane position. Based on the fingerprint feature vector, several groups of inbound target vehicles and outbound target vehicles are identified and associated.

Images