A method, system, device and medium for connecting real-time queuing trajectories of radar vision in intersection traffic scenes

Abstract

The present invention relates to a method, system, device and medium for connecting real-time queue trajectories of radar vision in a traffic scene at an intersection, and the method includes: determining that the trajectory is an entrance track or an exit track and performing coordinate conversion; classifying, updating or excluding the newly detected trajectory and the original trajectory; screening out the main and / or candidate trajectories that enter the intersection within a detection cycle, combining the entrance lane index and the conversion coordinates of the latest trajectory point of the trajectory to generate an exit trajectory set; determining the queue length and queue trajectory set of each lane of the entrance road; screening the main trajectory of the entrance road to generate a slow-moving trajectory set; operating based on the exit trajectory set, queue length, queue trajectory set and slow-moving trajectory set to achieve trajectory update and repair. The present invention greatly improves the accuracy and efficiency of queue detection through steps such as refined processing strategies, precise connection of trajectories, processing of slow-moving vehicles and accurate judgment of newly added queued vehicles.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a method, system, device and medium for connecting real-time queuing trajectories of radar vision in a traffic scene at an intersection. Background Art

[0002] In the current holographic digital road solution, millimeter wave radar and visual detection equipment are mainly used to achieve high-precision real-time tracking of road traffic targets. Especially in the scene of urban road intersections, the vehicle queue at the entrance is the key point of traffic control, and it is also the area that needs to be focused on when the holographic digital road is digitally mapped.

[0003] Due to the technical limitations of millimeter-wave radar's ability to recognize stationary targets, when motor vehicles are in a stationary queue, millimeter-wave radar is prone to problems such as target loss and prediction errors. This is mainly because the working principle of millimeter-wave radar focuses more on the detection of dynamic targets, and its ability to recognize stationary or slightly moving targets is relatively weak. On the other hand, although visual inspection equipment can provide richer information, in actual applications, due to the limitations of detection distance and the influence of factors such as large vehicle occlusion, visual equipment is also prone to missed detection, multiple detection, false alarms and other problems in complex queuing scenarios. These problems together lead to the lack of accuracy of digital roads in displaying real vehicle queuing scenarios.

[0004] In response to the above problems, the industry has taken some improvement measures. For example, by strengthening the chip performance of millimeter wave radar, using fixed frequency, directional transmitting and receiving antennas, and integrating intermediate frequency demodulation, signal amplification and digital processing functions, the detection ability of micro-movement and stationary targets can be improved. At the same time, visual inspection equipment also adds vertical illumination detection to more accurately capture the stationary, slow-moving, and starting behaviors of queued targets at the entrance.

[0005] However, despite these improvements, the current holographic trajectory of the queue area still has significant problems. First, when dealing with complex queue conditions on the entrance road, millimeter-wave radar still finds it difficult to avoid problems such as missed detection, false detection, and false alarm. Especially in congested sections, the trajectory detection error is large. Secondly, although the visual detection method can provide more sophisticated detection effects, its deployment cost is high, and additional equipment needs to be installed. In addition, the detection range is limited. The monitoring range of direct-illumination visual equipment is relatively small, and the factor of occlusion by large vehicles cannot be completely eliminated. These problems make it impossible for the visual detection method to achieve ideal and stable detection effects in practical applications. Finally, due to the detection errors of the visual trajectory and the radar trajectory, the connection between the two is often abnormal, and a large number of jumps are prone to occur, which further affects the accuracy and reliability of the holographic digital road. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a method, system, device and medium for connecting the real-time queuing trajectory of radar vision in a traffic scene at an intersection, which solves the technical problems that both millimeter-wave radar and visual detection equipment have technical limitations such as insufficient recognition accuracy, susceptibility to interference and abnormal connection of the trajectories of the two in queuing scenes at complex urban road intersections.

[0008] (II) Technical solution

[0009] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0010] In a first aspect, an embodiment of the present invention provides a method for connecting radar real-time queue trajectories in a traffic scene at an intersection, comprising:

[0011] According to the trajectory change information, determine whether the trajectory within the queuing detection range is an entrance track or an exit track, and perform coordinate conversion based on a reference line to obtain the conversion coordinates of each trajectory point;

[0012] Classify, update or exclude the newly detected trajectories and original trajectories within the queuing range as the main and / or candidate trajectories;

[0013] Screen out the main and / or candidate trajectories that enter the intersection within a set detection period, and generate an exit trajectory set by combining the entry lane index to which the trajectory belongs before entering the intersection and the transformed coordinates of the latest trajectory point of the screened trajectory;

[0014] Determine the queue length and queue trajectory set of each lane of the entrance road;

[0015] According to the number of new trajectory points and the queue status within the set detection time, the classified main trajectory of the entrance road is screened to generate a slow-moving trajectory set;

[0016] Based on the exit trajectory set, the queue length, the queue trajectory set and the slow-moving trajectory set, at least one operation of vehicle trajectory connection, queue vehicle addition and slow-moving vehicle processing is performed to achieve trajectory update and repair.

[0017] Optionally, determining the trajectory within the queuing detection range as an entrance track or an exit track according to the trajectory change information, and performing coordinate transformation based on a reference line to obtain the transformation coordinates of each trajectory point includes:

[0018] Determine a queue detection range;

[0019] Taking the stop line of the opposite exit road of the target entrance as the reference line, the coordinates of each trajectory are transformed by calculating the vertical distance between each trajectory coordinate and the stop line to obtain the transformed coordinates of each trajectory point;

[0020] Get the positive or negative change of the trajectory point conversion coordinates within the queuing detection range;

[0021] The trajectory type of the trajectory is determined to be an entrance track or an exit track according to the sign of the change in the conversion coordinates of the trajectory points.

[0022] Optionally, classifying, updating or excluding the newly detected tracks and original tracks within the queuing range as the main and / or candidate tracks includes:

[0023] Detect new original trajectories that appear within the queue detection range;

[0024] Determine whether the latest track point of the newly detected track is within the queue detection range, and determine whether the track direction is the import direction or the export direction;

[0025] If the trajectory direction is the import direction and is generated within the queue detection range, the trajectory is classified as an import candidate trajectory and is not output as the main trajectory;

[0026] If the trajectory direction is the import direction and is not generated within the queue detection range, the trajectory is classified as the main import trajectory, and the default queue status initial value is not queued;

[0027] If the trajectory direction is the exit direction and is inside the intersection, the trajectory is classified as a candidate trajectory and waits to connect with the main trajectory inside the intersection;

[0028] If the trajectory direction is the exit direction and is on the exit path, the trajectory is classified as the main trajectory and output;

[0029] For the original main track of the entrance road, determine the parking state of the original main track of the entrance road;

[0030] If the parking status of the original main track of the entrance road is a queued vehicle, it will remain as the main track of the entrance road;

[0031] If the parking status of the original entrance road main trajectory is non-queuing, but the latest trajectory point of the trajectory is within the queuing detection range and the direction is the entrance direction, it will remain as the entrance road main trajectory;

[0032] For the original candidate trajectory of the entrance road, if the latest trajectory point newly generated within a specific period of time is within the queue detection range and the trajectory direction is the import direction, it will be kept as the candidate trajectory of the entrance road;

[0033] The remaining trajectories are excluded from the scope of the main / candidate trajectories of the import channel.

[0034] Optionally, the main and / or candidate trajectories entering the intersection within a set detection period are screened out, and the exit trajectory set is generated by combining the import lane index to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory point of the screened trajectory, including:

[0035] Determine the testing cycle and testing period;

[0036] Preliminarily screening out the main and / or candidate trajectories in the import lane that have new trajectory points in the previous detection cycle and detection period;

[0037] For the initially screened trajectories, identify the transformed coordinates of the last trajectory point in the previous detection cycle;

[0038] If the negative square of the transformed coordinates of the last trajectory point in the previous detection cycle x -2 Greater than or equal to the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road X 1, and the negative first power of the transformed coordinates of the last track point in the current detection cycle x -1 Less than the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road X 1, then output the filtered trajectory;

[0039] The index of the entrance lane before the selected trajectory enters the intersection is determined according to the coordinates of the last trajectory point in the previous detection cycle;

[0040] According to the conversion coordinates of the latest trajectory points of the imported lane index and the filtered trajectory, a trajectory ID set of each imported lane entering the intersection within a set time period is generated and arranged, and then an exit trajectory set is obtained.

[0041] Optionally, determining the queue length and queue trajectory set of each lane of the entrance lane includes:

[0042] An initial queue length is set for each entry lane based on the vertical distance between the stop line of the entry lane and the stop line of the opposite exit lane;

[0043] Create an empty queue trajectory set to store the id of the main trajectory vehicle;

[0044] Add the queued vehicles to the queued trajectory set in sequence according to the value of their latest output point to generate a queued trajectory set;

[0045] Based on the added queue trajectory set and the set headway parameters, the queue length of each lane is calculated;

[0046] Output the queue length and queue trajectory set of each lane in the last detection cycle.

[0047] Optionally, according to the number of new trajectory points that appear within a set detection time and the queue status, the classified main trajectory of the entrance road is screened to generate a slow-moving trajectory set including:

[0048] Create an empty slow-moving trajectory set to store the id of the main trajectory vehicle;

[0049] If the main trajectory of the import channel does not generate a new trajectory point within the detection time and is in a non-queuing state, it is determined that the main trajectory has jumped;

[0050] Move the main trajectory that has jumped into the slow-moving trajectory set of the corresponding lane;

[0051] Traverse all the main tracks of the entrance road in the current detection cycle, generate a slow-moving trajectory set of the current detection cycle for each main trajectory, and output the latest trajectory points of the queued vehicles.

[0052] Optionally, based on the exit trajectory set, the queue length, the queue trajectory set, and the slow-moving trajectory set, performing at least one operation of vehicle trajectory connection, queue vehicle addition, and slow-moving vehicle processing to achieve trajectory update and repair includes:

[0053] Taking the lane as the basic unit, based on the exit trajectory set, queue length and queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and queue length are updated;

[0054] For the specific main trajectory and the slow-moving trajectory in the slow-moving trajectory set that are in the non-queuing state and generate new trajectory points, determine the lane to which they belong and form a candidate queuing trajectory set, evaluate the queuing state of the candidate queuing trajectory set one by one according to the queuing judgment conditions, form a newly added queuing vehicle set, and based on the new queuing trajectory set and the newly added queuing vehicle set, process the outgoing vehicles, configure the slow-moving queue, and update the queuing order, and update the queuing trajectory set and the queue length;

[0055] After at least one of the following operations is performed on all lanes of the target entrance road: vehicle trajectory connection, outgoing vehicle deletion, queued vehicle addition, and slow-moving vehicle deletion, the trajectory update and repair is completed.

[0056] Optionally, taking the lane as a basic unit, based on the exit trajectory set, the queue length and the queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and the queue length are updated, including:

[0057] In the vehicle trajectory connection process, for each import lane, it is determined whether the exit trajectory set is an empty set;

[0058] If the exit trajectory set is an empty set, the queue length and queue trajectory set are updated to the latest data at the end time of the previous detection cycle, and the connection of the import lane is terminated, and the connection process of other import lanes is started;

[0059] If the exit trajectory set is not an empty set, determine whether the queue trajectory set at the end of the previous detection cycle is an empty set;

[0060] If the queue trajectory set at the end of the previous detection cycle is an empty set, for all the entry trajectories in the exit trajectory set, the main trajectory and the candidate trajectory are regarded as the main trajectory, and the queue length and the queue trajectory set are updated to the latest data at the end of the previous detection cycle, and the connection of the import lane is ended, and the connection process of other import lanes is started;

[0061] If the queue trajectory set at the end of the previous detection cycle is not an empty set, a one-to-one match is performed between the exit trajectory set and the queue trajectory set at the end of the previous detection cycle according to the vehicle ID, and the last matched vehicle is taken as the reference vehicle. The reference vehicle and the vehicles before the reference vehicle in the queue trajectory set at the end of the previous detection cycle are considered to be exits, and the trajectories of the successfully matched vehicles are connected, and the remaining trajectories are connected in the queue order or output according to the logic of the main trajectory;

[0062] After the connection is completed, the queued vehicle trajectory that is considered to have left is deleted from the queued trajectory set, a new queued trajectory set is obtained, and it is determined whether the new queued trajectory set is an empty set;

[0063] If the new queue trajectory set is an empty set, there is no need to update the trajectory coordinates of the queued vehicles, only the queue length is updated;

[0064] If the new queue trajectory set is not an empty set, the latest trajectory points of the queue vehicles are output according to the number of remaining queue vehicles, and the queue length is updated.

[0065] Optionally, for a specific main trajectory in a non-queuing state and generating a new trajectory point and a slow-moving trajectory in a slow-moving trajectory set, determining the lane to which it belongs and forming a candidate queuing trajectory set, evaluating the queuing state of the candidate queuing trajectory set one by one according to the queuing judgment condition, forming a newly added queuing vehicle set, and performing outgoing vehicle processing, slow-moving queue configuration, and queuing sequence update based on the new queuing trajectory set and the newly added queuing vehicle set, and updating the queuing trajectory set and the queuing length, including:

[0066] Analyze the main trajectory in the set of specific main trajectories and slow-moving trajectories that are in a non-queue state in the entrance channel and generate new trajectory points during the detection period and whose last trajectory points are within the queue detection range;

[0067] The lane where the main trajectory is located is determined according to the coordinates of the last trajectory point of the main trajectory in the specific main trajectory and slow-moving trajectory set, and the size of the converted coordinates of the new trajectory point of the main trajectory in the specific main trajectory and slow-moving trajectory set is combined to form a candidate queuing trajectory set;

[0068] According to the queuing judgment condition, it is judged in turn whether the main trajectory in the candidate queuing trajectory set is a queuing vehicle;

[0069] The queue state of the main track is determined according to the judgment conditions; wherein the judgment conditions include: the speed of the last two track points is lower than the queue speed threshold, the x-axis value of the coordinate of the last track point is ≤ the queue length of the corresponding import lane, or the main track is in the slow-moving set list;

[0070] If the main track in the candidate queuing track set is not a queuing vehicle, then the track point is output according to the main track logic, and / or the queuing state of the next main track in the candidate queuing track set is judged until all main tracks are traversed or it is determined that there is no new queuing vehicle;

[0071] If the main track in the candidate queuing track set is a queuing vehicle, then the traversal of the subsequent main tracks in the candidate queuing track set is terminated, and the current main track and all subsequent main tracks are all judged as queuing vehicles and set to a queuing state to form a newly added queuing vehicle set;

[0072] If the new queue trajectory set and the newly added queue vehicle set are not empty sets, then analyze the transformation coordinates of the latest trajectory point in the newly added queue vehicle set X min Relationship with import lane queue length;

[0073] like X min If the queue length is less than 0, the main trajectory in the newly added queue vehicle set is set to the queue state, the first several vehicles in the new queue trajectory set are considered as outgoing vehicles, and the outgoing vehicles are deleted from the new queue trajectory set and the newly added queue vehicle set;

[0074] like X min If the queue length minus the queue length is greater than a specific value, the newly added queued vehicles are regarded as a slow-moving queue, and all main trajectories of the slow-moving queue are set to a non-queuing state and configured to move forward at a fixed speed;

[0075] like X min Subtract the queue length which is not less than 0 and X min If the queue length minus the length is not greater than a specific value, the main trajectories in the newly added queue vehicle set are arranged from small to large according to the value of the conversion coordinates, added to the queue trajectory set of the import lane, and the queue trajectory set and queue length are updated.

[0076] In a second aspect, an embodiment of the present invention provides a system for connecting radar real-time queue trajectories in a traffic scene at an intersection, comprising:

[0077] A coordinate conversion module is used to determine whether the trajectory within the queuing detection range is an entrance track or an exit track according to the trajectory change information, and to perform coordinate conversion based on a reference line to obtain the conversion coordinates of each trajectory point;

[0078] A trajectory processing module, used to classify, update or exclude the newly detected trajectories and original trajectories within the queuing range as the main and / or candidate trajectories;

[0079] An exit information output module is used to screen out the main and / or candidate trajectories that enter the intersection within a set detection period, and generate an exit trajectory set by combining the import lane index to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory point of the screened trajectory;

[0080] The queue information output module is used to determine the queue length and queue trajectory set of each lane of the entrance road;

[0081] The slow-moving information output module is used to filter the main trajectories of the entrance obtained by classification according to the number of new trajectory points appearing and the queue status within the set detection time to generate a slow-moving trajectory set;

[0082] The track updating and repairing module is used to perform at least one operation of vehicle track connection, queue vehicle addition and slow-moving vehicle processing based on the exit track set, queue length, queue track set and slow-moving track set to achieve track updating and repair.

[0083] In a third aspect, an embodiment of the present invention provides a device for connecting real-time radar queue trajectories in a traffic scene at an intersection, comprising: at least one database; and a memory communicatively connected to the at least one database; wherein the memory stores instructions executable by the at least one database, and the instructions are executed by the at least one database so that the at least one database can execute the method for connecting real-time radar queue trajectories in a traffic scene at an intersection as described above.

[0084] In a fourth aspect, an embodiment of the present invention provides a computer-readable medium having computer executable instructions stored thereon, and when the executable instructions are executed by a processor, the method for connecting the real-time radar queuing trajectory in the intersection traffic scene as described above is implemented.

[0085] (III) Beneficial effects

[0086] The beneficial effects of the present invention are:

[0087] First, the present invention realizes the unified processing of trajectory data generated by multiple types of detectors by determining the trajectory direction, defining the queue detection range, and performing coordinate conversion. This step not only simplifies the subsequent data processing process, but also improves the accuracy and comparability of the data. Through coordinate conversion, it can be applied to multiple types of detectors such as radar and vision at the same time, realizing seamless switching and improving the applicability and flexibility of the algorithm.

[0088] Secondly, the present invention adopts refined processing strategies in judging the main trajectory and candidate trajectory of the entrance lane, the trajectory of the outgoing vehicle, and calculating the queue length and queue trajectory set. These strategies cover all driving behaviors of vehicles at the entrance lane, including entering, exiting, queuing, slow movement, etc., and realize comprehensive monitoring of the traffic scene at the intersection. This step can update the status of the main trajectory and candidate trajectory in real time, and can accurately track the driving trajectory of each vehicle. At the same time, by screening the trajectory of the outgoing vehicle and judging the import lane it belongs to before entering the intersection, accurate data support is provided for subsequent queue optimization.

[0089] Next, the present invention realizes real-time monitoring and optimization of the queue situation at the intersection by calculating the queue length and queue trajectory set of the import lane. This step ensures that the algorithm can accurately calculate the queue situation of each import lane, so as to realize dynamic adjustment and optimization of the queue situation at the intersection.

[0090] In addition, the present invention further improves the accuracy and reliability of the algorithm by determining the slow-moving vehicle set, updating the slow-moving vehicle set and predicting the trajectory, and judging the newly added queued vehicles. These steps can initialize the slow-moving vehicle set list and update the slow-moving vehicle set according to the state of the main trajectory. These processes effectively solve the abnormal effects such as vehicle crushing, vehicle interruption, and false alarm vehicles caused by detection accuracy problems.

[0091] Therefore, the real-time queue trajectory connection method of radar vision in intersection traffic scenarios of the present invention not only greatly improves the accuracy and efficiency of queue detection through refined processing strategies, precise connection of trajectories, processing of slow-moving vehicles, and accurate judgment of newly added queued vehicles, but also enhances the algorithm's adaptability to various traffic scenarios, providing strong technical support for intersection traffic management. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 A flow chart of a method for connecting radar-visual real-time queuing trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0093] Figure 2 A specific flow chart of step S1 of the method for connecting radar-visual real-time queue trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0094] Figure 3 A specific flow chart of step S2 of the method for connecting radar-visual real-time queue trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0095] Figure 4 A specific flow chart of step S3 of the method for connecting radar real-time queue trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0096] Figure 5 A specific flow chart of step S4 of the method for connecting radar-visual real-time queue trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0097] Figure 6 A specific flow chart of step S5 of the method for connecting radar-visual real-time queue trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0098] Figure 7 A specific flow chart of step S6 of the method for connecting radar-visual real-time queue trajectories in a traffic scene at an intersection provided by an embodiment of the present invention;

[0099] Figure 8 A schematic diagram of the overall process of the method for connecting radar real-time queuing trajectories in an intersection traffic scenario provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0100] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0101] Original trajectory: Real-time trajectory data generated at a certain frequency after a roadside detector (millimeter wave radar, visual detector) detects a target on the road. A trajectory data needs to include the most basic information, namely: trajectory id (i.e. obj_id), detection time, trajectory point coordinates (latitude and longitude, coordinates of the detector's own coordinate system, etc.).

[0102] Main trajectory: After receiving the original trajectory, the algorithm processes the original trajectory in a series of ways (including screening, connection, prediction, etc.) to output the main trajectory, which is the trajectory that is finally mapped to the front-end digital map. The main trajectory is composed of segments of original trajectories connected together, and also includes: trajectory id (i.e. vehicle_id), detection time, and trajectory point coordinates.

[0103] Candidate tracks: After the original tracks meet certain conditions, some of them will become candidate tracks. Candidate tracks can only be converted into main tracks and output to the digital map after being connected with existing main tracks.

[0104] Entering vehicles: Vehicles entering the queue detection range are entering vehicles.

[0105] Queuing vehicles: Vehicles that are stationary within the queuing detection range are queuing vehicles.

[0106] Exiting vehicles: Vehicles that exit the queue detection range and enter the intersection are exiting vehicles.

[0107] Slow-moving vehicle: A vehicle that is within the queue detection range, moves at a lower speed, and whose driving behavior meets certain conditions is a slow-moving vehicle.

[0108] Entrance lane and exit lane: For this intersection, the lanes where vehicles are traveling toward the intersection are called entrance lanes, and the lanes where vehicles are traveling away from the intersection are called exit lanes.

[0109] Convert coordinates x :The vertical distance from the stop line of the exit road opposite to the import is x Coordinates, in meters. x The coordinate is the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road. X 1. Original trajectory x The coordinates are the vertical distance between the coordinates of the trajectory point and the stop line of the exit road opposite the import.

[0110] like Figure 1 As shown, a method for connecting real-time radar queue trajectories in an intersection traffic scene proposed in an embodiment of the present invention includes: determining, according to trajectory change information, whether a trajectory within a queue detection range is an entrance lane trajectory or an exit lane trajectory, and performing coordinate conversion with a reference line to obtain conversion coordinates of each trajectory point; classifying, updating or excluding the main and / or candidate trajectories of the entrance lane for the newly detected trajectory and the original trajectory within the queue range; screening out the main and / or candidate trajectories that enter the intersection within a set detection period, and generating an exit trajectory set by combining the index of the entrance lane to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory points of the screened trajectory; determining the queue length and queue trajectory set of each lane of the entrance lane; screening the classified entrance lane main trajectory according to the number of new trajectory points and the queue status within the set detection time to generate a slow-moving trajectory set; performing at least one operation of vehicle trajectory connection, queue vehicle addition and slow-moving vehicle processing based on the exit trajectory set, queue length, queue trajectory set and slow-moving trajectory set to achieve trajectory update and repair.

[0111] First, the present invention realizes the unified processing of trajectory data generated by multiple types of detectors by determining the trajectory direction, defining the queue detection range, and performing coordinate conversion. This step not only simplifies the subsequent data processing process, but also improves the accuracy and comparability of the data. Through coordinate conversion, it can be applied to multiple types of detectors such as radar and vision at the same time, realizing seamless switching and improving the applicability and flexibility of the algorithm.

[0112] Secondly, the present invention adopts refined processing strategies in judging the main trajectory and candidate trajectory of the entrance lane, the trajectory of the outgoing vehicle, and calculating the queue length and queue trajectory set. These strategies cover all driving behaviors of vehicles at the entrance lane, including entering, exiting, queuing, slow movement, etc., and realize comprehensive monitoring of the traffic scene at the intersection. This step can update the status of the main trajectory and candidate trajectory in real time, and can accurately track the driving trajectory of each vehicle. At the same time, by screening the trajectory of the outgoing vehicle and judging the import lane it belongs to before entering the intersection, accurate data support is provided for subsequent queue optimization.

[0113] Next, the present invention realizes real-time monitoring and optimization of the queue situation at the intersection by calculating the queue length and queue trajectory set of the import lane. This step ensures that the algorithm can accurately calculate the queue situation of each import lane, so as to realize dynamic adjustment and optimization of the queue situation at the intersection.

[0114] In addition, the present invention further improves the accuracy and reliability of the algorithm by determining the slow-moving vehicle set, updating the slow-moving vehicle set and predicting the trajectory, and judging the newly added queued vehicles. These steps can initialize the slow-moving vehicle set list and update the slow-moving vehicle set according to the state of the main trajectory. These processes effectively solve the abnormal effects such as vehicle crushing, vehicle interruption, and false alarm vehicles caused by detection accuracy problems.

[0115] Therefore, the real-time queue trajectory connection method of radar vision in intersection traffic scenarios of the present invention not only greatly improves the accuracy and efficiency of queue detection through refined processing strategies, precise connection of trajectories, processing of slow-moving vehicles, and accurate judgment of newly added queued vehicles, but also enhances the algorithm's adaptability to various traffic scenarios, providing strong technical support for intersection traffic management.

[0116] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0117] Specifically, an embodiment of the present invention provides a method for connecting real-time queuing trajectories of radar vision in a traffic scene at an intersection, which includes:

[0118] S1. Determine whether the trajectory within the queuing detection range is an entrance track or an exit track according to the trajectory change information, and perform coordinate transformation based on a reference line to obtain the transformation coordinates of each trajectory point.

[0119] Furthermore, if Figure 2 As shown, step S1 includes:

[0120] S11. Determine a queuing detection range.

[0121] S12, using the stop line of the exit road opposite to the target entrance as the reference line, by calculating the vertical distance between each trajectory coordinate and the stop line, coordinate transformation is performed on each trajectory coordinate of the trajectory to obtain the transformed coordinates of each trajectory point. The purpose of this step is to unify the trajectory data generated by different detectors into the same coordinate system for subsequent processing and analysis.

[0122] S13, obtaining the positive or negative change of the conversion coordinates of the trajectory points within the queuing detection range.

[0123] S14. Determine whether the trajectory type of the trajectory is an entrance track or an exit track according to the sign of the change of the trajectory point of the trajectory.

[0124] In one embodiment, the detection device X 1 Taking the import inspection as an example, the queue inspection range is defined as [ x 0, x 1], for the target trajectory generated within the road range obj i d , select the stop line of the opposite exit road corresponding to the target entrance as the baseline. In order to obtain the relative position relationship between each trajectory and this baseline, all trajectory coordinates are transformed. The result after transformation, that is, the vertical distance between each trajectory coordinate and the stop line, is defined as the new transformed coordinate of the trajectory x This step ensures that the positions of different trajectories relative to the exit stop line can be evaluated and analyzed with a unified metric, providing a basis for subsequent queue detection and analysis.

[0125] Next, transform the coordinates according to the trajectory points of the trajectory x The change in the sign of is judged as the entrance track or exit track. Specifically, if the transformed coordinates of the track point x If it increases with time, it is identified as the entrance track; on the contrary, if it decreases, it is identified as the exit track. In addition, the direction of the main track should be determined in accordance with the direction of the current fusion. obj i d The indicated directions remained consistent to ensure accuracy and consistency in trajectory analysis.

[0126] S2. Classify, update or exclude the newly detected trajectories and original trajectories within the queuing range as the main and / or candidate trajectories.

[0127] Furthermore, if Figure 3 As shown, step S2 includes:

[0128] S21. Detect new original trajectories within the queue detection range.

[0129] S22, judging whether the latest track point of the newly detected track is within the queuing detection range, and determining whether the track direction is an import direction or an export direction.

[0130] S23a. If the trajectory direction is the import direction and is generated within the queue detection range, the trajectory is classified as an import candidate trajectory and is not output as a main trajectory. If the trajectory direction is the import direction and is not generated within the queue detection range, the trajectory is classified as an import main trajectory and the default queue status initial value is not queued.

[0131] S23b, if the trajectory direction is the exit direction and is inside the intersection, the trajectory is classified as a candidate trajectory and waits to connect with the main trajectory inside the intersection; if the trajectory direction is the exit direction and is on the exit road, the trajectory is classified as a main trajectory and output.

[0132] S24. For the original main trajectory of the entrance road, determine the parking status of the original main trajectory of the entrance road; if the parking status of the original main trajectory of the entrance road is a queuing vehicle, then keep it as the main trajectory of the entrance road; if the parking status of the original main trajectory of the entrance road is non-queuing, but the latest trajectory point of the trajectory is within the queuing detection range and the direction is the import direction, then keep it as the main trajectory of the entrance road.

[0133] S25. For the original candidate trajectory of the entrance road, if the latest trajectory point newly generated within a specific time period is within the queuing detection range and the trajectory direction is the entrance direction, it is maintained as the candidate trajectory of the entrance road; the remaining trajectories are excluded from the main / candidate trajectory range of the entrance road.

[0134] In one embodiment, real-time monitoring and detection are performed to determine whether a new original track appears in the set queue detection area. For each newly detected track, the position of its latest track point is analyzed to determine whether it is within the queue detection range. At the same time, the direction of travel of the track is determined, and it is distinguished as an import direction or an export direction.

[0135] For the detected in the queue detection range [ x 0, x 1], if the latest trajectory point of the trajectory is within the queue detection range and the trajectory direction is the import direction, the main trajectory and candidate trajectory judgment begins:

[0136] If the trajectory is within the queue detection range [ x 0, x 1], the trajectory is not output as the main trajectory, but as a candidate trajectory for the entrance.

[0137] The remaining entrance track trajectories that enter the queue detection range are output as the main entrance track trajectories, and the default queue state initial value is S q =0, indicating that there is no queue at present.

[0138] If the trajectory is heading towards the exit, then according to the exit road logic, it is determined whether it is located at the exit road or inside the intersection. If it is inside the intersection, it is used as a candidate trajectory waiting to connect with the main trajectory inside the intersection. If it is an exit road, it is output as the main trajectory.

[0139] For the original main trajectory of the entrance road and the candidate trajectory, if the parking state of the main trajectory of the entrance road is S q =1, that is, it is a queued vehicle, then it is still the main trajectory of the entrance road; if the parking state of the main trajectory of the entrance road is S q = 0, if the latest trajectory point of the trajectory is within the queue detection range and the trajectory direction is the import direction, it is still the main trajectory of the import road; for the original candidate trajectory, if the latest trajectory point of the trajectory (must be within the detection period Δt If a new trajectory (newly generated during the time period) is within the queue detection range and the trajectory direction is the import direction, it is still a candidate trajectory for the import road. The rest are excluded from the main / candidate trajectory range of the import road.

[0140] S3, screening out the main and / or candidate trajectories entering the intersection within a set detection period, and combining the import lane index to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory point of the screened trajectory to generate an exit trajectory set.

[0141] Furthermore, if Figure 4 As shown, step S3 includes:

[0142] S31. Determine a detection cycle and a detection period.

[0143] S32: Preliminarily screen out the main and / or candidate trajectories in the import lane that have new trajectory points in the previous detection cycle and detection period.

[0144] S33: For the initially screened trajectory, identify the conversion coordinates of the last trajectory point in the previous detection cycle.

[0145] S34, if the negative square of the conversion coordinates of the last trajectory point in the previous detection cycle x -2 Greater than or equal to the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road X 1, and the negative first power of the transformed coordinates of the last track point in the current detection cycle x -1Less than the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road X 1, then the filtered trajectory is output.

[0146] S35, judging the import lane index before the selected trajectory enters the intersection according to the coordinates of the last trajectory point in the previous detection cycle.

[0147] S36. Generate and arrange the trajectory ID set of each import lane entering the intersection within a set time period according to the conversion coordinates of the latest trajectory point of the trajectory obtained by the import lane index and screening, and then obtain the exit trajectory set.

[0148] In one embodiment, the detection period is determined Δt is 600ms, and the current time is t , the current detection period is [ t - Δ t, t ]. The main task of this step is to determine the outgoing vehicle, that is, to identify Δt Cross the stop line X 1 and enter the intersection range, that is, the outgoing vehicle trajectory. The specific method of screening the outgoing vehicle trajectory is as follows:

[0149] In the previous detection cycle [ t -2 Δt, t-Δt ] All original trajectories in the import lane (excluding non-motorized vehicle lanes) as candidate trajectories and main trajectories, and in Δt There are new trajectory points within the time period (i.e. no jump).

[0150] The trajectory in the previous detection cycle [ t -2 Δt, t-Δt ]The transformed coordinates of the last trajectory point in x satisfy x -2 ≥ X 1 and the trajectory is in the current detection cycle [ t - Δt, t ]The transformed coordinates of the last trajectory point in x satisfy x -1 < X 1.

[0151] In the screening obtained Δt After setting the trajectory of entering the intersection within the time period, according to its [ t -2 Δt, t-Δt ] time period to determine which import lane the trajectory belongs to before entering the intersection, and generate each import lane in [ t - Δt, t] The set of trajectory IDs that enter the intersection during the period, and the conversion coordinates of the latest trajectory point of the trajectory x -1 Arrange the values ​​from small to large to obtain the import lane k Collection List , which is the latest Δt The set of exit trajectories within the time period .

[0152] S4. Determine the queue length and queue trajectory set of each lane of the entrance lane.

[0153] Furthermore, if Figure 5 As shown, step S4 includes:

[0154] S41. An initial queue length is set for each import lane based on the vertical distance between the stop line of the import lane and the stop line of the opposite exit lane.

[0155] S42: Create an empty queue trajectory set to store the ID of the main trajectory vehicle.

[0156] S43. Add the queued vehicles to the queued trajectory set in sequence according to the values ​​of their latest output points to generate a queued trajectory set.

[0157] S44. Calculate the queue length of each lane based on the added queue trajectory set and the set vehicle headway parameter.

[0158] S45. Output the queue length and queue trajectory set of each lane in the previous detection cycle.

[0159] In a specific embodiment, first determine the queue length of each lane of the import lane, and initialize the import lane k The queue length is At the same time, define a queue trajectory set as ,in The id of the main track vehicle, the set list is sorted according to the conversion coordinates of its latest output point x The values ​​of are sorted from small to large. The initial value of the queue list is an empty set.

[0160] Next, set the configuration parameters, including the headway distance. ΔX , which is used to calculate the actual occupied length of each lane when vehicles queue up. Assume that the current time is t , considering the lanes detected in the previous detection cycle k The queue situation of the previous period is recorded as [ t -2 Δt, t-Δt ] Detected lanes k The set of queue trajectories is , the queue length is .

[0161] S5. According to the number of new trajectory points appearing within the set detection time and the queue status, the classified main trajectory of the entrance road is screened to generate a slow-moving trajectory set.

[0162] Furthermore, if Figure 6 As shown, step S5 includes:

[0163] S51. Create an empty slow-moving trajectory set for storing the ID of the main trajectory vehicle.

[0164] S52: If the main trajectory of the inlet track does not generate a new trajectory point within the detection time and is in a non-queuing state, it is determined that the main trajectory has a jump.

[0165] S53: Move the main trajectory that has jumped into the slow-moving trajectory set of the corresponding lane.

[0166] S54, traverse all the main tracks of the entrance road in the current detection cycle, generate a slow-moving trajectory set of the current detection cycle for each main trajectory, and output the latest trajectory point of the queued vehicle.

[0167] In one embodiment, the slow motion trajectory set is defined as , where the elements are transformed according to the coordinates of their latest output point x The values ​​are sorted from small to large. During initialization, the initial value of the slow-moving trajectory set is an empty set. Set the configuration parameter headway to ΔX . Let the current time be t , the detection cycle of the previous period [ t -2 Δt, t-Δt ] Detected lanes k The slow moving trajectory set is .

[0168] Next, the main trajectory of the entrance road obtained by classification is analyzed. Δt No new points are generated within the period, and the queue status S q = 0 (which means it is not currently in any other queue state, but may be included in the list of slow-moving trajectory sets), then the main trajectory is judged to have jumped, and the trajectory prediction phase is entered, and it is moved into the slow-moving set of the corresponding lane. S k ( t - Δt ).

[0169] Then, all the main trajectories of the current round of import channels are traversed, and for each main trajectory, a set of slow-moving trajectories in the current detection cycle is generated. S k (t ), and output the latest trajectory points of these queued vehicles as follows:

[0170] ;

[0171] The timestamp of the trajectory point is t , V 1 is the calibration parameter, the fixed speed of the slow-moving vehicle.

[0172] S6. Based on the exit trajectory set, the queue length, the queue trajectory set, and the slow-moving trajectory set, at least one operation of vehicle trajectory connection, queue vehicle addition, and slow-moving vehicle processing is performed to update and repair the trajectory.

[0173] Furthermore, if Figure 7 As shown, step S6 includes:

[0174] S61. Taking the lane as the basic unit, based on the exit trajectory set, the queue length and the queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and the queue length are updated.

[0175] Further, step S61 includes:

[0176] In the vehicle trajectory connection process, for each import lane, it is determined whether the exit trajectory set is an empty set.

[0177] If the exit trajectory set is an empty set, the queue length and queue trajectory set are updated to the latest data at the end time of the previous detection cycle, and the connection of the import lane is ended, and the connection processing of other import lanes is started.

[0178] If the exit trajectory set is not an empty set, determine whether the queue trajectory set at the end of the previous detection cycle is an empty set.

[0179] If the queue trajectory set at the end of the previous detection cycle is an empty set, for all the entry trajectories in the exit trajectory set, the main trajectory and the candidate trajectory are regarded as the main trajectory, and the queue length and the queue trajectory set are updated to the latest data at the end of the previous detection cycle, and the connection of the import lane is ended, and the connection processing of other import lanes is started.

[0180] If the queue trajectory set at the end of the previous detection cycle is not an empty set, a one-to-one match is performed between the exit trajectory set and the queue trajectory set at the end of the previous detection cycle according to the vehicle ID, and the last matched vehicle is taken as the reference vehicle. The reference vehicle and the vehicles before the reference vehicle in the queue trajectory set at the end of the previous detection cycle are considered to have exited, and the trajectories of the successfully matched vehicles are connected, and the remaining trajectories are connected in the queue order or output according to the logic of the main trajectory.

[0181] After the connection is completed, the queued vehicle trajectory that is considered to have left is deleted from the queued trajectory set, a new queued trajectory set is obtained, and it is determined whether the new queued trajectory set is an empty set.

[0182] If the new queue trajectory set is an empty set, there is no need to update the trajectory coordinates of the queued vehicles, only the queue length is updated.

[0183] If the new queue trajectory set is not an empty set, the latest trajectory points of the queue vehicles are output according to the number of remaining queue vehicles, and the queue length is updated.

[0184] In one embodiment, the import lane k For example, the import lane has been obtained k The set of queue trajectories , the last detection cycle[ t -2 Δt, t-Δt ] Detected lanes k The set of queue trajectories is , the queue length is Specifically, the vehicle trajectory connection process is as follows:

[0185] like is an empty set, that is Δt No entry into the intersection during the time period k Lane vehicles, then update the import lane k The set of queue trajectories is , queue length , the latest output point of all queued vehicle trajectories in the set is the output point of the previous period (i.e., stationary), the connection of this lane ends, and the connection screening of other import lanes begins; if Not an empty set, get The transformed coordinates of the latest trajectory points of all the trajectories entering the intersection x The maximum value among the values ​​of X 2.

[0186] like Not an empty set, determine the queue trajectory set at the end of the previous detection cycle Is it an empty set? If Is an empty set, that is, the import lane kNo queues of vehicles. All the trajectories entering the intersection in the main track are output according to the logic of the main track as usual if the trajectory is already the main track; if the trajectory is a candidate track, the candidate track is judged as the main track and output according to the logic of the main track to update the import lane. k The set of queue trajectories is , queue length .

[0187] End the connection of this lane and enter the connection with other import lanes of this entrance.

[0188] like Not an empty set: and First, match the original track IDs one by one, taking the last vehicle in the pair as the benchmark (conversion x Coordinates are the largest), and the queue will be The vehicle and the vehicles before it are considered to have left; for example, the set of exit trajectories of a lane is , the queue trajectory set is , then 2 pairs are successfully paired. As the benchmark, the queue trajectory set All are driven out. The successfully matched vehicles are connected according to their IDs. The remaining trajectory and The remaining tracks in the queue are connected one by one in the order of queueing. After the connection is completed, if There are remaining trajectories, for If the remaining track in the main track is already the main track, it will be output according to the logic of the main track as usual.

[0189] After the connection is completed, update the import lane k The queue vehicle list and queue length are deleted from the queue vehicle list to obtain a new queue trajectory set. . It is known that the import stop line x The coordinates are X 1. The headroom is ΔX , The transformed coordinates of the latest trajectory points of all the trajectories entering the intersection x The maximum value among the values X 2.

[0190] like If it is an empty set, that is, all the queued vehicles have entered the intersection, and there are no queued vehicles in the import lane at present, there is no need to update the trajectory coordinates of the queued vehicles, and update the import lane k The queue length is .

[0191] like is not an empty set, let the number of remaining unconnected queue vehicles be n , sorted by Then output the latest trajectory points of these queued vehicles:

[0192] The trajectory points (1≤ i ≤ n )for:

[0193] ;

[0194] The output trajectory point timestamp is t， Import lane k The queue length is .

[0195] End of import lane k According to the above process, the other lanes of the entrance road are connected with the outgoing vehicles and the queued vehicles again. If all lanes of the entrance road have been connected, the update is completed.

[0196] S62. For the specific main trajectory that is in a non-queuing state and generates new trajectory points and the slow-moving trajectory in the slow-moving trajectory set, determine the lane to which it belongs and form a candidate queuing trajectory set, evaluate the queuing state of the candidate queuing trajectory set one by one according to the queuing judgment condition, form a newly added queuing vehicle set, and based on the new queuing trajectory set and the newly added queuing vehicle set, perform outgoing vehicle processing, slow-moving queue configuration, and queuing order update, and update the queuing trajectory set and the queue length.

[0197] Further, step S62 includes:

[0198] The main trajectory in the set of specific main trajectories and slow-moving trajectories that are in a non-queuing state in the entrance channel, generate new trajectory points during the detection period, and whose last trajectory points are within the queuing detection range are analyzed.

[0199] The lane where the main trajectory is located is determined according to the coordinates of the last trajectory point of the main trajectory in the specific main trajectory and slow-moving trajectory set, and the candidate queuing trajectory set is formed by combining the size of the converted coordinates of the new trajectory point of the main trajectory in the specific main trajectory and slow-moving trajectory set.

[0200] According to the queuing judgment conditions, it is judged in turn whether the main trajectory in the candidate queuing trajectory set is a queuing vehicle.

[0201] The queue status of the main track is determined according to the judgment conditions; the judgment conditions include: the speed of the last two track points is lower than the queue speed threshold, the coordinates of the last track point x The value of the axis ≤ the queue length of the corresponding import lane or the main trajectory is in the slow-moving set list.

[0202] If the main track in the candidate queuing track set is not a queuing vehicle, the track point is output according to the main track logic, and / or the queuing status of the next main track in the candidate queuing track set is judged until all main tracks are traversed or it is determined that there is no new queuing vehicle.

[0203] If the main trajectory in the candidate queuing trajectory set is a queuing vehicle, the traversal of subsequent main trajectories in the candidate queuing trajectory set is terminated, and the current main trajectory and all subsequent main trajectories are all judged as queuing vehicles and set to a queuing state to form a newly added queuing vehicle set.

[0204] If the new queue trajectory set and the newly added queue vehicle set are not empty sets, then analyze the transformation coordinates of the latest trajectory point in the newly added queue vehicle set X min Relationship with import lane queue length.

[0205] like X min If the queue length is less than 0, the main trajectory in the newly added queue vehicle set is set to the queue state, the first several vehicles in the new queue trajectory set are regarded as outgoing vehicles, and the outgoing vehicles are deleted from the new queue trajectory set and the newly added queue vehicle set.

[0206] like X min If the queue length minus the length of the queue is greater than a specific value, the newly added queued vehicle set is regarded as a slow-moving queue, all main trajectories of the slow-moving queue are set to a non-queuing state, and are configured to move forward at a fixed speed.

[0207] like X min Subtract the queue length which is not less than 0 and X min If the queue length minus the length is not greater than a specific value, the main trajectories in the newly added queue vehicle set are arranged from small to large according to the value of the conversion coordinates, added to the queue trajectory set of the import lane, and the queue trajectory set and queue length are updated.

[0208] In another specific embodiment, for all queued S q =0, in Δt The main trajectory (including the main trajectory in the slow-motion set list) within the detection range is judged by generating a new trajectory point within the time. The specific steps are as follows:

[0209] 1) Determine the lane where the main trajectory is located, determine its lane number based on the coordinates of the last trajectory point of the trajectory, obtain the candidate queue vehicle set for each import lane, and calculate the conversion coordinates of the new trajectory points of each main trajectory in the set. x , arranged from largest to smallest, with the import lanek For example, the candidate queuing vehicle sequence set for this detection cycle is , and the set has n Main track.

[0210] 2) Candidate queuing trajectory set for the import lane The main trajectory in the sequence set is determined according to the sorting order to determine whether the trajectory is a queue vehicle. Let the current one be the first m The judgment of (1≤m≤n) main trajectories, the import lane is known k The current new queue trajectory set is , the queue length is , determine the import lane k The queuing state of the main trajectory of is determined if any of the following conditions are met:

[0211] a. The coordinates of the last two trajectory points of the original trajectory of the main trajectory are ( x 1, y 1, t 1) and ( x 2, y 2, t 2), if the speed between the coordinates of the last two trajectory points is lower than the configured queue speed threshold, e.g. ,in For configuration parameters, the queue speed threshold is set, and it is judged as a queue vehicle.

[0212] b. The coordinates of the final track point of the original track of the main track are ( x 1, y 1, t 1) If , it is judged as a queue vehicle.

[0213] c. The main track is in the slow-motion set list.

[0214] 3) Next, process non-queue vehicles and continue traversal: If the main trajectory is not a queue vehicle, its trajectory points are output as usual according to the main trajectory logic. m +1≤ n , then return to step 2) for the first m +1 main track to judge the queue status, otherwise, end the collection Traversal judgment of all main trajectories in the import lane k There are no new queued vehicles during this cycle of detection, and new queued vehicles in other lanes are detected.

[0215] 4) Then, process the queued vehicles and end the traversal: If the main track is judged as a queued vehicle, the set is also terminated. The traversal judgment of the main trajectory in m Main tracks and collections All main tracks sorted after this main track are judged as queued vehicles, forming the import lane of this detection cycle. k The collection of newly added queued vehicles , set the queue status of all main tracks in this collection to ,in .

[0216] 5) If the new queue trajectory set Not empty, and a new queue vehicle collection is added If it is not empty, then record middle The transformed coordinates of the latest trajectory point x for X min ,like , then set The queue status of all main tracks in is ,in ,See (before) (Round up) The vehicle is the exiting vehicle, and its current detection cycle is set to the same trajectory point ( y value unchanged), and then change it from If these vehicles are located in the slow-moving trajectory set In the same , and then go to 7); if , then Treat it as a slow-moving queue and set the queue status of all its main tracks to , the speed is ,in , the queue moves at a fixed speed V 1, then go to 7); if and , go to 6).

[0217] 6) The main trajectory of the queue in x The values ​​are arranged from small to large, and then added to the import lane k The list of queued vehicles In the above example, we can obtain the set of queued vehicles detected in this detection cycle. , let the latest trajectory point of the original trajectory of the queue vehicle be , then the latest trajectory point output to the front end is :

[0218] ;

[0219] in g for The number of trajectories in .

[0220] Update import lanes k The length of the queued vehicles in this detection cycle is .

[0221] 7) Return to step 1) to complete the algorithm for adding queuing vehicles in other lanes of the target entrance. After traversing all the entrance lanes, the queuing optimization algorithm for the target entrance in this detection cycle is completed.

[0222] S63: After performing at least one of the following operations on all lanes of the target entrance lane, namely, connecting vehicle trajectories, deleting outgoing vehicles, adding queued vehicles, and deleting slow-moving vehicles, the trajectory update and repair are completed.

[0223] In addition, an embodiment of the present invention provides a system for connecting radar real-time queue trajectories in a traffic scene at an intersection, including:

[0224] The coordinate conversion module is used to determine whether the trajectory of the queuing detection range is an entrance track or an exit track according to the trajectory change information, and to perform coordinate conversion based on a reference line to obtain the conversion coordinates of each trajectory point.

[0225] The trajectory processing module is used to classify, update or exclude the newly detected trajectories and original trajectories within the queuing range as the main and / or candidate trajectories.

[0226] The exit information output module is used to screen out the main and / or candidate trajectories that enter the intersection within a set detection period, and generate an exit trajectory set by combining the import lane index to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory point of the screened trajectory.

[0227] The queue information output module is used to determine the queue length and queue trajectory set of each lane of the entrance road.

[0228] The slow-moving information output module is used to filter the classified main trajectories of the entrance road according to the number of new trajectory points appearing and the queue status within the set detection time to generate a slow-moving trajectory set.

[0229] The track updating and repairing module is used to perform at least one operation of vehicle track connection, queue vehicle addition and slow-moving vehicle processing based on the exit track set, queue length, queue track set and slow-moving track set to achieve track updating and repair.

[0230] Furthermore, an embodiment of the present invention provides a device for connecting real-time radar queue trajectories in intersection traffic scenarios, including: at least one database; and a memory communicatively connected to at least one database; wherein the memory stores instructions that can be executed by at least one database, and the instructions are executed by at least one database so that at least one database can execute the method for connecting real-time radar queue trajectories in intersection traffic scenarios as described above.

[0231] Once the instructions stored in the memory are executed by the database, they can drive the database to execute a complete set of real-time queue trajectory connection methods based on radar and visual information in intersection traffic scenarios according to the preset logic. This method can efficiently integrate data from radar and visual sensors to achieve accurate monitoring and analysis of vehicle dynamics and queue conditions at intersections, thereby providing real-time and accurate data support for intelligent traffic management systems, optimizing traffic flow and improving intersection traffic efficiency.

[0232] Furthermore, an embodiment of the present invention provides a computer-readable medium having computer executable instructions stored thereon, and when the executable instructions are executed by a processor, the method for connecting the real-time radar queuing trajectories in the intersection traffic scene as described above is implemented.

[0233] Specifically, once these executable instructions are executed, they can realize a real-time queue trajectory connection method that integrates radar and visual information in intersection traffic scenes. This means that through the computing power of the processor, combined with the instructions stored on the medium, the system can collect, process and analyze data from radar and visual sensors in real time, and then accurately identify and track the queues of vehicles at the intersection and their driving trajectories.

[0234] In summary, the embodiments of the present invention provide a method, system, device and medium for connecting real-time queuing trajectories of radar vision in a traffic scene at an intersection, such as Figure 8 As shown, first, the core step is to accurately distinguish whether the trajectory within the queuing detection range belongs to the entrance or exit lane based on the trajectory change information, and implement coordinate transformation through a baseline to obtain the accurate transformation coordinates of each trajectory point.

[0235] Next, the newly detected trajectories and original trajectories within the queue range are carefully classified, including the distinction, update or exclusion of the main trajectory of the entrance road and the candidate trajectory when necessary. On this basis, the main and candidate trajectories that enter the intersection within a set detection period are screened out, and the exit trajectory set is generated by combining the import lane index before the trajectory enters the intersection and the conversion coordinates of the latest trajectory point. Subsequently, the queue length and queue trajectory set of each lane of the entrance road are determined, and the slow-moving trajectory in the main trajectory of the entrance road is further screened out according to the number of new trajectory points and the queue status to form a slow-moving trajectory set.

[0236] Taking the lane as the basic analysis unit, the outgoing trajectory set, queue length and queue trajectory set are used to achieve the effective connection between the outgoing vehicles and the queue vehicles in each import lane, output the trajectory point conversion coordinates of the outgoing vehicles and the pre-queue vehicles, and dynamically update the queue trajectory set and queue length. For specific main trajectories and slow-moving trajectories that are not in a queue state but generate new trajectory points, the lanes to which they belong will be determined to form a candidate queue trajectory set, and they will be evaluated one by one according to the queue judgment conditions to finally form a new queue vehicle set. Based on these sets, outgoing vehicle processing, slow-moving queue configuration and queue order update are carried out, and the queue trajectory set and queue length are updated again.

[0237] After this series of operations, necessary operations such as vehicle trajectory connection, outgoing vehicle deletion, queue vehicle addition and slow-moving vehicle deletion will be performed on all lanes of the target entrance to ensure the update and repair of the trajectory.

[0238] Specifically: First, the adaptation of multiple types of detectors is a highlight of the present invention. The algorithm is cleverly designed, and can unify and convert the coordinate system of the detection data generated by various detectors (such as radar, video, etc.) to ensure the consistency and comparability of the data. At the same time, in response to the different challenges faced by radar and video in queue detection, the algorithm provides targeted optimization strategies, thereby effectively improving the accuracy and reliability of detection. Secondly, decision-making is the main and detection is the auxiliary is the core idea of ​​the present invention. In view of the accuracy problem of current queue detection, this algorithm abandons the traditional practice of relying solely on detection data, but makes a comprehensive decision on the driving behavior of the vehicle through the fusion and logical judgment between multiple categories. The detection information of the vehicle itself is used as an auxiliary means to help the algorithm grasp the vehicle behavior more accurately and further improve the accuracy and robustness of the decision. Furthermore, trajectory prediction and connection are the key links for the present invention to achieve seamless connection. For the jump trajectory, the algorithm can make intelligent predictions and judgments based on its current state and the behavior of the adjacent vehicles to ensure the continuity and accuracy of the trajectory. At the same time, for the trajectories detected between different devices, the algorithm can also effectively connect to ensure that the trajectory data of the entire traffic scene is complete and consistent.

[0239] The advantages of the present invention are mainly reflected in the following aspects:

[0240] Unified trajectory coordinate system: By performing corresponding analysis and processing on the radar trajectory, the coordinate system of the trajectory coordinates is unified, so that the algorithm can be simultaneously applied to the queue trajectories of multiple types of detectors such as radar and vision, achieving seamless switching and greatly improving the versatility and practicality of the algorithm.

[0241] Comprehensive coverage of vehicle driving behaviors: The present invention covers all driving behaviors of vehicles at the entrance, including driving in, driving out, queuing, slow moving, etc. When a vehicle is lost, the algorithm can timely make up for the queue; when the vehicle starts driving, it can make predictions and connections; when the vehicle track stops due to false alarms, it can make predictions and repairs; when the vehicle moves slowly, it can make predictions and repairs. This series of measures effectively prevents abnormal effects such as vehicle crushing, vehicle interruption, and false alarm vehicles caused by detection accuracy problems.

[0242] Compensating for detector deficiencies: The present invention effectively compensates for the deficiencies and differences of various detectors in detecting stationary vehicles by processing the radar trajectory in an integrated manner. The algorithm eliminates the detection differences between trajectory data of different detectors, improves the connection effect between trajectories of multiple data sources, and provides strong support for the intelligent management of intersection traffic scenes.

[0243] Since the system / device described in the above embodiments of the present invention is a system / device used to implement the method of the above embodiments of the present invention, a person skilled in the art can understand the specific structure and deformation of the system / device based on the method described in the above embodiments of the present invention, and thus will not be described in detail here. All systems / devices used in the method of the above embodiments of the present invention belong to the scope of protection of the present invention.

[0244] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0245] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0246] It should be noted that the word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising a number of different components and by means of a suitably programmed computer. Among the several means listed, several of these means may be embodied by the same hardware. The use of the words first, second, third, etc., is for convenience of expression only and does not indicate any order. These words may be understood as part of the name of the component.

[0247] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0248] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments after knowing the basic creative concept. Therefore, the technical solution should be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0249] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the technical solution of the present invention and its equivalent technology, the present invention should also include these modifications and variations.

Claims

1. A method for connecting real-time queuing trajectories of radar vision in a traffic scene at an intersection, characterized in that: include: According to the trajectory change information, determine whether the trajectory within the queuing detection range is an entrance track or an exit track, and perform coordinate conversion based on a reference line to obtain the conversion coordinates of each trajectory point; Classify, update or exclude the newly detected trajectories and original trajectories within the queuing range as the main and / or candidate trajectories; Screen out the main and / or candidate trajectories that enter the intersection within a set detection period, and generate an exit trajectory set by combining the entry lane index to which the trajectory belongs before entering the intersection and the transformed coordinates of the latest trajectory point of the screened trajectory; Determine the queue length and queue trajectory set of each lane of the entrance road; According to the number of new trajectory points and the queue status within the set detection time, the classified main trajectory of the entrance road is screened to generate a slow-moving trajectory set; Based on the exit trajectory set, the queue length, the queue trajectory set and the slow-moving trajectory set, at least one operation of vehicle trajectory connection, queue vehicle addition and slow-moving vehicle processing is performed to achieve trajectory update and repair; Wherein, based on the exit trajectory set, the queue length, the queue trajectory set and the slow-moving trajectory set, at least one operation of vehicle trajectory connection, queue vehicle addition and slow-moving vehicle processing is performed to achieve trajectory update and repair, including: Taking the lane as the basic unit, based on the exit trajectory set, queue length and queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and queue length are updated; For the specific main trajectory and the slow-moving trajectory in the slow-moving trajectory set that are in the non-queuing state and generate new trajectory points, determine the lane to which they belong and form a candidate queuing trajectory set, evaluate the queuing state of the candidate queuing trajectory set one by one according to the queuing judgment conditions, form a newly added queuing vehicle set, and based on the new queuing trajectory set and the newly added queuing vehicle set, process the outgoing vehicles, configure the slow-moving queue, and update the queuing order, and update the queuing trajectory set and the queue length; After at least one of the following operations is performed on all lanes of the target entrance lane, namely, vehicle track connection, outgoing vehicle deletion, queue vehicle addition, and slow-moving vehicle deletion, the track update and repair is completed; Taking the lane as the basic unit, based on the exit trajectory set, queue length and queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and queue length are updated, including: In the vehicle trajectory connection process, for each import lane, it is determined whether the exit trajectory set is an empty set; If the exit trajectory set is an empty set, the queue length and queue trajectory set are updated to the latest data at the end time of the previous detection cycle, and the connection of the import lane is terminated, and the connection process of other import lanes is started; If the exit trajectory set is not an empty set, determine whether the queue trajectory set at the end of the previous detection cycle is an empty set; If the queue trajectory set at the end of the previous detection cycle is an empty set, for all the entry trajectories in the exit trajectory set, the main trajectory and the candidate trajectory are regarded as the main trajectory, and the queue length and the queue trajectory set are updated to the latest data at the end of the previous detection cycle, and the connection of the import lane is ended, and the connection process of other import lanes is started; If the queue trajectory set at the end of the previous detection cycle is not an empty set, a one-to-one match is performed between the exit trajectory set and the queue trajectory set at the end of the previous detection cycle according to the vehicle ID, and the last matched vehicle is taken as the reference vehicle. The reference vehicle and the vehicles before the reference vehicle in the queue trajectory set at the end of the previous detection cycle are considered to be exits, and the trajectories of the successfully matched vehicles are connected, and the remaining trajectories are connected in the queue order or output according to the logic of the main trajectory; After the connection is completed, the queued vehicle trajectory that is considered to have left is deleted from the queued trajectory set, a new queued trajectory set is obtained, and it is determined whether the new queued trajectory set is an empty set; If the new queue trajectory set is an empty set, there is no need to update the trajectory coordinates of the queued vehicles, only the queue length is updated; If the new queue trajectory set is not an empty set, the latest trajectory points of the queue vehicles are output according to the number of remaining queue vehicles, and the queue length is updated.

2. The method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as claimed in claim 1 is characterized in that: According to the trajectory change information, the trajectory within the queue detection range is determined to be the entrance track or the exit track, and the coordinates are transformed with a reference line to obtain the transformation coordinates of each trajectory point, including: Determine a queue detection range; Taking the stop line of the opposite exit road of the target entrance as the reference line, the coordinates of each trajectory are transformed by calculating the vertical distance between each trajectory coordinate and the stop line to obtain the transformed coordinates of each trajectory point; Get the positive or negative change of the trajectory point conversion coordinates within the queuing detection range; The trajectory type of the trajectory is determined to be an entrance track or an exit track according to the sign of the change in the conversion coordinates of the trajectory points.

3. The method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as claimed in claim 1 is characterized in that: Classifying, updating or excluding the newly detected tracks and original tracks in the queue range as the main and / or candidate tracks includes: Detect new original trajectories that appear within the queue detection range; Determine whether the latest track point of the newly detected track is within the queue detection range, and determine whether the track direction is the import direction or the export direction; If the trajectory direction is the import direction and is generated within the queue detection range, the trajectory is classified as an import candidate trajectory and is not output as the main trajectory; If the trajectory direction is the import direction and is not generated within the queue detection range, the trajectory is classified as the main import trajectory, and the default queue status initial value is not queued; If the trajectory direction is the exit direction and is inside the intersection, the trajectory is classified as a candidate trajectory and waits to connect with the main trajectory inside the intersection; If the trajectory direction is the exit direction and is on the exit path, the trajectory is classified as the main trajectory and output; For the original main track of the entrance road, determine the parking state of the original main track of the entrance road; If the parking status of the original main track of the entrance road is a queued vehicle, it will remain as the main track of the entrance road; If the parking status of the original entrance road main trajectory is non-queuing, but the latest trajectory point of the trajectory is within the queuing detection range and the direction is the entrance direction, it will remain as the entrance road main trajectory; For the original candidate trajectory of the entrance road, if the latest trajectory point newly generated within a specific period of time is within the queue detection range and the trajectory direction is the import direction, it will be kept as the candidate trajectory of the entrance road; The remaining trajectories are excluded from the scope of the main / candidate trajectories of the import channel.

4. The method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as claimed in claim 1 is characterized in that: The main and / or candidate trajectories entering the intersection within a set detection period are screened out, and the exit trajectory set is generated by combining the import lane index to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory point of the screened trajectory, including: Determine the testing cycle and testing period; Preliminarily screening out the main and / or candidate trajectories in the import lane that have new trajectory points in the previous detection cycle and detection period; For the initially screened trajectories, identify the transformed coordinates of the last trajectory point in the previous detection cycle; If the negative square of the transformed coordinates of the last trajectory point in the previous detection cycle x -2 Greater than or equal to the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road X 1, and the negative first power of the transformed coordinates of the last track point in the current detection cycle x -1 Less than the vertical distance between the stop line of the entrance road and the stop line of the opposite exit road X 1, then output the filtered trajectory; The index of the entrance lane before the selected trajectory enters the intersection is determined according to the coordinates of the last trajectory point in the previous detection cycle; According to the conversion coordinates of the latest trajectory points of the imported lane index and the filtered trajectory, a trajectory ID set of each imported lane entering the intersection within a set time period is generated and arranged, and then an exit trajectory set is obtained.

5. The method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as claimed in claim 1 is characterized in that: Determine the queue length and queue trajectory set of each lane of the entrance road including: An initial queue length is set for each entry lane based on the vertical distance between the stop line of the entry lane and the stop line of the opposite exit lane; Create an empty queue trajectory set to store the id of the main trajectory vehicle; Add the queued vehicles to the queued trajectory set in sequence according to the value of their latest output point to generate a queued trajectory set; Based on the added queue trajectory set and the set headway parameters, the queue length of each lane is calculated; Output the queue length and queue trajectory set of each lane in the last detection cycle.

6. The method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as claimed in claim 1, characterized in that: According to the number of new trajectory points and the queue status within the set detection time, the classified main trajectory of the entrance road is screened to generate a slow-moving trajectory set including: Create an empty slow-moving trajectory set to store the id of the main trajectory vehicle; If the main trajectory of the import channel does not generate a new trajectory point within the detection time and is in a non-queuing state, it is determined that the main trajectory has jumped; Move the main trajectory that has jumped into the slow-moving trajectory set of the corresponding lane; Traverse all the main tracks of the entrance road in the current detection cycle, generate a slow-moving trajectory set of the current detection cycle for each main trajectory, and output the latest trajectory points of the queued vehicles.

7. The method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as claimed in claim 1, characterized in that: For the specific main trajectory and the slow-moving trajectory in the slow-moving trajectory set that are not in the queue state and generate new trajectory points, determine the lane to which they belong and form a candidate queue trajectory set. According to the queue judgment conditions, the queue state of the candidate queue trajectory set is evaluated one by one to form a newly added queue vehicle set. Based on the new queue trajectory set and the newly added queue vehicle set, the exit vehicle processing, slow-moving queue configuration and queue order update are performed, and the queue trajectory set and queue length are updated, including: Analyze the main trajectory in the set of specific main trajectories and slow-moving trajectories that are in a non-queue state in the entrance channel and generate new trajectory points during the detection period and whose last trajectory points are within the queue detection range; The lane where the main trajectory is located is determined according to the coordinates of the last trajectory point of the main trajectory in the specific main trajectory and slow-moving trajectory set, and the candidate queuing trajectory set is formed by combining the size of the converted coordinates of the new trajectory point of the main trajectory in the specific main trajectory and slow-moving trajectory set; According to the queuing judgment condition, it is judged in turn whether the main trajectory in the candidate queuing trajectory set is a queuing vehicle; The queue state of the main track is determined according to the judgment conditions; wherein the judgment conditions include: the speed of the last two track points is lower than the queue speed threshold, the x-axis value of the coordinate of the last track point is ≤ the queue length of the corresponding import lane, or the main track is in the slow-moving set list; If the main track in the candidate queuing track set is not a queuing vehicle, the track point is output according to the main track logic, and / or the queuing state of the next main track in the candidate queuing track set is judged until all main tracks are traversed or it is determined that there is no new queuing vehicle; If the main track in the candidate queuing track set is a queuing vehicle, then the traversal of the subsequent main tracks in the candidate queuing track set is terminated, and the current main track and all subsequent main tracks are all judged as queuing vehicles and set to a queuing state to form a newly added queuing vehicle set; If the new queue trajectory set and the newly added queue vehicle set are not empty sets, then analyze the transformation coordinates of the latest trajectory point in the newly added queue vehicle set X min Relationship with import lane queue length; like X min If the queue length is less than 0, the main trajectory in the newly added queue vehicle set is set to the queue state, the first several vehicles in the new queue trajectory set are considered as outgoing vehicles, and the outgoing vehicles are deleted from the new queue trajectory set and the newly added queue vehicle set; like X min If the queue length minus the queue length is greater than a specific value, the newly added queued vehicles are regarded as a slow-moving queue, and all main trajectories of the slow-moving queue are set to a non-queuing state and configured to move forward at a fixed speed; like X min Subtract the queue length which is not less than 0 and X min If the queue length minus the length is not greater than a specific value, the main trajectories in the newly added queue vehicle set are arranged from small to large according to the value of the conversion coordinates, added to the queue trajectory set of the import lane, and the queue trajectory set and queue length are updated.

8. A system for connecting real-time queuing trajectories of radar vision in a traffic scene at an intersection, characterized in that: include: A coordinate conversion module is used to determine whether the trajectory within the queuing detection range is an entrance track or an exit track according to the trajectory change information, and to perform coordinate conversion based on a reference line to obtain the conversion coordinates of each trajectory point; A trajectory processing module, used to classify, update or exclude the newly detected trajectories and original trajectories within the queuing range as the main and / or candidate trajectories; An exit information output module is used to screen out the main and / or candidate trajectories that enter the intersection within a set detection period, and generate an exit trajectory set by combining the import lane index to which the trajectory belongs before entering the intersection and the conversion coordinates of the latest trajectory point of the screened trajectory; The queue information output module is used to determine the queue length and queue trajectory set of each lane of the entrance road; The slow-moving information output module is used to filter the main trajectories of the entrance obtained by classification according to the number of new trajectory points appearing and the queue status within the set detection time to generate a slow-moving trajectory set; A trajectory updating and repairing module, for performing at least one of vehicle trajectory connection, queue vehicle addition, and slow-moving vehicle processing based on the exit trajectory set, queue length, queue trajectory set, and slow-moving trajectory set, so as to achieve trajectory updating and repairing; Wherein, based on the exit trajectory set, the queue length, the queue trajectory set and the slow-moving trajectory set, at least one operation of vehicle trajectory connection, queue vehicle addition and slow-moving vehicle processing is performed to achieve trajectory update and repair, including: Taking the lane as the basic unit, based on the exit trajectory set, queue length and queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and queue length are updated; For the specific main trajectory and the slow-moving trajectory in the slow-moving trajectory set that are in the non-queuing state and generate new trajectory points, determine the lane to which they belong and form a candidate queuing trajectory set, evaluate the queuing state of the candidate queuing trajectory set one by one according to the queuing judgment conditions, form a newly added queuing vehicle set, and based on the new queuing trajectory set and the newly added queuing vehicle set, process the outgoing vehicles, configure the slow-moving queue, and update the queuing order, and update the queuing trajectory set and the queue length; After at least one of the following operations is performed on all lanes of the target entrance lane, namely, vehicle track connection, outgoing vehicle deletion, queue vehicle addition, and slow-moving vehicle deletion, the track update and repair is completed; Taking the lane as the basic unit, based on the exit trajectory set, queue length and queue trajectory set, the exiting vehicles in each entrance lane are connected with the queueing vehicles, the trajectory point conversion coordinates of the exiting vehicles and the pre-queuing vehicles are output, and the queue trajectory set and queue length are updated, including: In the vehicle trajectory connection process, for each import lane, it is determined whether the exit trajectory set is an empty set; If the exit trajectory set is an empty set, the queue length and queue trajectory set are updated to the latest data at the end time of the previous detection cycle, and the connection of the import lane is terminated, and the connection process of other import lanes is started; If the exit trajectory set is not an empty set, determine whether the queue trajectory set at the end of the previous detection cycle is an empty set; If the queue trajectory set at the end of the previous detection cycle is an empty set, for all the entry trajectories in the exit trajectory set, the main trajectory and the candidate trajectory are regarded as the main trajectory, and the queue length and the queue trajectory set are updated to the latest data at the end of the previous detection cycle, and the connection of the import lane is ended, and the connection process of other import lanes is started; If the queue trajectory set at the end of the previous detection cycle is not an empty set, a one-to-one match is performed between the exit trajectory set and the queue trajectory set at the end of the previous detection cycle according to the vehicle ID, and the last matched vehicle is taken as the reference vehicle. The reference vehicle and the vehicles before the reference vehicle in the queue trajectory set at the end of the previous detection cycle are considered to be exits, and the trajectories of the successfully matched vehicles are connected, and the remaining trajectories are connected in the queue order or output according to the logic of the main trajectory; After the connection is completed, the queued vehicle trajectory that is considered to have left is deleted from the queued trajectory set, a new queued trajectory set is obtained, and it is determined whether the new queued trajectory set is an empty set; If the new queue trajectory set is an empty set, there is no need to update the trajectory coordinates of the queued vehicles, only the queue length is updated; If the new queue trajectory set is not an empty set, the latest trajectory points of the queue vehicles are output according to the number of remaining queue vehicles, and the queue length is updated.

9. A device for connecting the real-time queuing track of radar vision in a traffic scene at an intersection, characterized in that: include: at least one database; And a memory communicatively connected to at least one database; wherein the memory stores instructions executable by at least one database, and the instructions are executed by at least one database so that at least one database can execute the method for connecting real-time radar queuing trajectories in intersection traffic scenarios as described in any one of claims 1-7.

10. A computer-readable medium having computer-executable instructions stored thereon, characterized in that: When the executable instructions are executed by the processor, the method for connecting the real-time queuing trajectory of radar vision in the intersection traffic scene as described in any one of claims 1-7 is implemented.

Images