A track connection optimization method for a multi-radar detection area of a road section

Abstract

The application discloses a kind of track connection optimization methods of section multi-radar detection area, overcome the single radar detection range not full in prior art, cannot obtain the complete trajectory of vehicle on intermediate section, and the trajectory obtained is prone to jump problem, comprising the following steps: S1: selecting radar layout point, define radar connection area;S2: obtain the target trajectory set detected by radar, the target trajectory set is filtered, and the target trajectory set after deduplication is obtained;S3: whether the target trajectory set after deduplication is multi-radar connection area, according to the judgment result carries out the same radar trajectory connection or multi-radar trajectory connection;S4: traverse all target trajectories, analyze the remaining candidate trajectory, and obtain the complete trajectory of section.The same radar trajectory connection and multi-radar trajectory connection are carried out by arranging radar in intermediate section, the complete trajectory information of vehicle in intermediate section is obtained, high accuracy, fast matching, and jump is effectively prevented simultaneously.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a method for optimizing trajectory connection in multiple radar detection areas of a road segment. Background Technology

[0002] Road traffic safety detection and management rely on advanced sensor detection technology. Currently, millimeter-wave radar terminal equipment is widely used in urban intersection environments for sensor technology in traffic infrastructure to acquire real-time traffic information. The effective detection range of millimeter-wave radar equipment installed at intersections is generally around 200 meters. Multiple radars installed in different locations can form a complete intersection detection area. However, for road sections between two intersections, due to the limitations of radar equipment detection capabilities, a single radar cannot complete the detection of the entire road section, thus making it impossible to obtain the complete trajectory of vehicles traveling in the middle section through a single radar detection. Furthermore, radar equipment cannot detect vehicles traveling at slow speeds, vehicles that are stationary, or vehicles obstructed by deployment height limitations, which may result in trajectory jumps and other problems.

[0003] For the detection of the middle section of the road, it is generally necessary to combine other detection technologies to fuse multi-source data to obtain the complete trajectory of the same vehicle. However, in the radar detection scenario, no specific solution has been given for trajectory connection based on multiple radar areas. Therefore, how to obtain the complete trajectory of vehicles traveling on the middle section based on multi-radar area detection information is an urgent problem to be solved in the application of radar in traffic scenarios. Summary of the Invention

[0004] The purpose of this invention is to overcome the problems in the prior art where the detection range of a single radar is incomplete for road sections between two intersections, making it impossible to obtain the complete trajectory of vehicles in the middle section, and the obtained trajectory is prone to jumps. This invention provides a method for optimizing the trajectory connection of multiple radar detection areas in a road section. By deploying radars in the middle section to connect the trajectory of the same radar and the trajectory of multiple radars, the complete trajectory information of vehicles in the middle section is obtained. This method has high accuracy, fast matching, and effectively prevents jumps.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for optimizing trajectory connection in multiple radar detection areas of a road segment, comprising the following steps:

[0006] S1: Select radar deployment locations and define the radar connection area;

[0007] S2: Obtain the set of target trajectories detected by the radar, filter the set of target trajectories, and obtain the set of target trajectories after deduplication;

[0008] S3: Determine whether the set of target trajectories after deduplication is a multi-radar connection area. If so, first connect the same radar trajectory, then filter candidate trajectories and connect the multi-radar trajectories. If not, determine whether a trajectory jump has occurred. If so, connect the same radar trajectory.

[0009] S4: Traverse all target trajectories, analyze the remaining candidate trajectories, and obtain the complete trajectory set of each target vehicle.

[0010] Based on radar detection capabilities, radar deployment points are selected along the road segment. Two radars should be deployed at the same radar deployment point to detect two opposing traffic flows. For two adjacent radar deployment points, the intermediate area is the multi-radar area, and the detection ranges of the two radars within this area must overlap. Deployment is completed along the entire road segment to achieve full coverage of the detection range. The method for connecting the trajectories of vehicles traveling in different directions within the multi-radar area is the same. This invention obtains complete trajectory information of vehicles in the intermediate road segment by deploying radars in the middle section and connecting both the same radar trajectory and the multi-radar trajectory. It achieves high accuracy, fast matching, and effectively prevents jumps in trajectory, while providing important reference for the detection and management of urban road traffic safety.

[0011] Preferably, step S2 is further expressed as follows:

[0012] S2.1: If radar X1 and radar X2 are millimeter-wave radars installed at two adjacent radar deployment points, obtain the target trajectory set detected by radar X1, filter the target trajectory set located in the multi-radar connection area Z0 at the current time, and deduplicate it as set A; filter the target trajectory set located in the same radar connection area Z1 at the current time, and deduplicate it as set B.

[0013] S2.2: Obtain the set of target trajectories detected by radar X2, filter the set of target trajectories located in the same radar junction area Z2 at the current time, and deduplicate them, denoted as set C.

[0014] If radar X1 and radar X2 are millimeter-wave radars installed at two adjacent radar deployment points, then the cross detection range of radar X1 and radar X2 is defined as the multi-radar connection area Z0, the area where radar X1 does not cross with radar X2 is defined as the same-radar connection area Z1, and the area where radar X2 does not cross with radar X1 is defined as the same-radar connection area Z2.

[0015] Preferably, step S3 is further expressed as follows:

[0016] For the multi-radar integration area: For each target trajectory in set A, perform trajectory integration with the same radar trajectory to obtain new trajectory points within the time period (tT,t] for each target trajectory, and update each target trajectory in set A; Filter and deduplicate the candidate trajectory set detected by radar X2: For each target trajectory, it must satisfy the following conditions: the total number of trajectory points within the time period (tT,t] must be greater than or equal to N1, and there must be trajectory points belonging to the multi-radar integration area Z0, resulting in set D; Perform multi-radar trajectory integration, and save the successfully integrated target trajectories in set G;

[0017] For areas not connected by multiple radars: the status of each target trajectory in sets B and C is determined: if no new trajectory points are generated within the time period (tT,t], the main trajectory changes and is connected to the radar trajectory, and the successfully connected target trajectory is saved in set G; if a new trajectory point is generated within (tT,t], the main trajectory does not change, and the corresponding detection status information of the main trajectory within (tT,t] is saved in set G.

[0018] N1 is the candidate trajectory filtering threshold 1, which is a configuration parameter; T represents the current time and the calculation period, which is defined as (tT, t] for the current period and (t-2*T, tT] for the previous period. This invention determines whether the area is a multi-radar connection region. If so, it performs trajectory connection with the same radar, then filters and deduplicates the candidate trajectory set to obtain set D, and then performs multi-radar trajectory connection again. If it is not a multi-radar region, it determines whether the trajectory jumps. If so, it performs trajectory connection with the same radar. Successfully connected candidate trajectory data is deleted, and the detection status information of the remaining target trajectories with a trajectory point count greater than or equal to N2 within D is stored in set D. Successfully connected trajectories are then added to set G, where N2 is the candidate trajectory filtering threshold 2, which is a configuration parameter. The target trajectory set in set G is the complete trajectory set of each target vehicle in that area.

[0019] Preferably, in step S3, the connection with the radar trajectory further includes:

[0020] A1: Based on the spatiotemporal range, a set of candidate trajectories for the main trajectory is obtained. Each target main trajectory is further filtered in the set of candidate trajectories based on trajectory features.

[0021] A2: Calculate the distance Δd between the first trajectory point of each target on the candidate trajectory and the last trajectory point of the main trajectory. Select the candidate trajectory with the smallest Δd and match it with the main trajectory. Assign the ID of the main trajectory to the candidate trajectory and end the connection between the main trajectory and the radar trajectory of the target. If a candidate trajectory meets the connection conditions of multiple target trajectories, connect it with the target trajectory with the smallest Δd. If the connection is successful, add a label mark_id to the target trajectory and mark mark_id=1. The connection with the radar trajectory is completed. If the connection fails, proceed to A3.

[0022] A3: If the latest point of the main trajectory is located in the multi-radar connection area Z0, add a label mark_id to the target trajectory and mark mark_id=0, and perform multi-radar area trajectory connection; if the latest point of the main trajectory is located in the same radar connection area Z1, add a virtual point.

[0023] Where Δd = ΔLon + ΔLat, ΔLon and ΔLat are the latitude and longitude distance differences between the first trajectory point and the latest trajectory point of each target in the candidate trajectory.

[0024] Preferably, step A1 is further expressed as:

[0025] A1.1: Based on the spatiotemporal range conditions: The trajectory points of candidate trajectory L1 and the trajectory points of the main trajectory L0 before the jump are detected by the same radar; the time of the first trajectory point of candidate trajectory L1 is later than the time of the last trajectory point of the main trajectory L0, and the time difference is within the range of (0,T1), where T1 is the trajectory matching time threshold 1, which is a configuration parameter with a value range of [0,+∞); the longitude and latitude difference between the first trajectory point of candidate trajectory L1 and the last trajectory point of the main trajectory L0 is within the range of ΔLon1 and ΔLat1, where ΔLon1 and ΔLat1 are configuration parameters with a value range of [0,0.01]; the number of trajectory points of candidate trajectory L1 within (tT,t] is greater than or equal to N3, where N3 is the candidate trajectory filtering threshold 3, which is a configuration parameter; based on the above four spatiotemporal range conditions, a set of candidate trajectories is obtained. If there are 0 candidate trajectories, the matching fails and proceeds to A3; otherwise, proceeds to A1.2.

[0026] A1.2: Select a set of candidate trajectories that satisfy the following conditions: the angle between the vector formed by the first point and the (1+N4)th point on the candidate trajectory and the vector formed by the last point and the (1+N4)th point from the end on the main trajectory is less than or equal to α1; and the angle between the vector formed by the last point on the main trajectory and the first point on the candidate vehicle trajectory and the vector formed by the last point and the (1+N4)th point from the end on the main trajectory is less than α2. If there are 0 candidate trajectories, the matching fails and proceeds to A3; otherwise, proceeds to A2.

[0027] For the same connecting target, the main trajectory is denoted as L0, and the candidate trajectory is denoted as L1.

[0028] Preferably, the multi-radar trajectory connection further includes:

[0029] B1: Based on the spatiotemporal range, a set of candidate trajectories for the main trajectory is obtained. Each target main trajectory is further filtered based on trajectory features in the obtained set of candidate trajectories for the main trajectory. Each target trajectory is traversed to obtain a set of candidate trajectories for each target main trajectory.

[0030] B2: Traverse each target trajectory to obtain a unique candidate trajectory for each target trajectory, and filter the candidate trajectory to find the unique main trajectory;

[0031] B3: If no suitable candidate trajectory is found, the search range is expanded to perform a second matching between the target's main trajectory and the candidate trajectory. If no candidate trajectory is found, virtual points are added to the target trajectory that fails to connect with the radar trajectory and whose latest point is located in the multi-radar connection area Z0. If a candidate trajectory is successfully matched, the spatiotemporal parameters of the latest trajectory point of the candidate trajectory are the spatiotemporal parameters of the latest trajectory point of the main trajectory within the marking period. The target ID of the main trajectory, the target ID of the connected candidate trajectory, and the detected state information are stored in set G.

[0032] The set of target trajectories in set G is the complete set of trajectories for all target vehicles in the current area. For target trajectories with mark_id = 1, the marking period is (tT, t], and for target trajectories with mark_id = 0, the marking period is (t-2*T, t+T). If no suitable candidate trajectory is found, the search range is appropriately expanded, and the coefficients ε are determined, letting T′2 = εT2, ΔLon′ = εΔLon2, ΔLat′ = εΔLat2, and ΔV′... X =εΔVx, α3 remains unchanged, where T2, ΔLon2, ΔLat2, ΔVx, and α3 are the time, distance, velocity, and angle configuration parameters for the first matching of the target's main trajectory and candidate trajectory during multi-radar integration, respectively. After expanding the determination coefficient ε, T′2, ΔLon2′, ΔLat2′, and ΔV′ are used. Xα3 is used as a parameter for secondary matching to perform secondary matching between the target's main trajectory and candidate trajectories. If a candidate trajectory is still not successfully matched, for target trajectories with mark_id=1 (successfully matched with the radar trajectory): retain the latest trajectory point obtained from the original radar trajectory connection, and similarly extract the target ID of this type of target trajectory, the target ID of the connected candidate trajectory, and the detected state information and store them in set G; for target trajectories with mark_id=0 (failed to match with the radar trajectory connection, but the latest point of the main trajectory is located in the multi-radar connection area Z0): add virtual points (the process is the same as adding virtual points in the radar trajectory connection), store the state information of the target trajectory with added virtual points in set G; delete the candidate trajectories that have been successfully connected from the candidate trajectory set.

[0033] Preferably, step B1 further includes:

[0034] Based on spatiotemporal range: For the same connecting target, obtain the spatiotemporal parameters of the trajectory points within the marked period in the target trajectory set A of the deduplicated multi-radar connecting area Z0 of the main trajectory. Iterate through the spatiotemporal parameters of each point on the candidate trajectory to perform point matching. If the time corresponding to the candidate trajectory point is within... Within, i represents whether the current connection is a connection with the same radar or multiple radars; if it is a connection with the same radar, i = 1; if it is a connection with multiple radars, i = 2. R1 represents the candidate trajectory points for matching. T represents the current detection time for this candidate trajectory point. i To match configuration parameters, or the difference between the longitude and latitude of candidate trajectory points and the longitude and latitude of main trajectory points in ΔLon i With ΔLat i If the location is within the specified range, it indicates a successful point matching, where ΔLon i With ΔLat i The distance threshold for matching parameters between the same radar / multiple radars is a configuration parameter;

[0035] Based on trajectory features: take the spatiotemporal parameters of the latest point and the 1+N4th point from the end of the candidate trajectory, as well as the spatiotemporal parameters of the latest point and the 1+N4th point from the end of the main trajectory, where N4 is the candidate trajectory screening threshold and 4 is a configuration parameter. If the main trajectory has two trajectory points within the marking period, take the spatiotemporal parameters of these two points; if the main trajectory has only one point within the marking period, take the spatiotemporal parameters of that point and the spatiotemporal parameters of the trajectory point with the longest time in the previous period of the target trajectory.

[0036] For the same connecting target, the main trajectory is denoted as L2. The spatiotemporal parameters of the trajectory points within the marked period in set A are obtained.

[0037] in The spatiotemporal parameter vector of the main trajectory point R1, This represents the current detection time for this trajectory point. and The coordinates of the trajectory point are latitude and longitude. and The coordinates of this trajectory point are in the radar coordinate system. The spatiotemporal parameters of each point on the candidate trajectory are iterated. Perform point matching, where Let R2 be the spatiotemporal parameter vector of the candidate trajectory point. This represents the current detection time for this trajectory point. and The coordinates of the trajectory point are latitude and longitude. and These are the radar coordinates of the trajectory point. The time corresponding to the main trajectory point being matched (when mark_id = 1, T) i =T2, when mark_id=0 T i =T2+1); ΔLat when mark_id=1 i =ΔLat2, ΔLat when mark_id=0 i =ΔLat2+0.0001.

[0038] As a preferred option, when filtering based on trajectory features, constraints are also included:

[0039] For target trajectories that are successfully matched with radar trajectories or fail to be matched with radar trajectories but whose latest point is located in the multi-radar connection area Z0, the following directional angle constraint condition is satisfied: the angle between the vector formed by the first point and the first (1+N4)th point on the candidate trajectory and the vector formed by the last point and the last (1+N4)th point on the main trajectory is less than or equal to α, where α1 and α3 are the trajectory matching angle thresholds 1 and 3, respectively, and both have a value range of [0, 360].

[0040] For a target trajectory that successfully matches the radar trajectory, the velocity constraint condition is satisfied: the velocity V is calculated using the X coordinates of the 1st point and the (1+N4th)th point on the candidate trajectory. R The velocity V is calculated using the X-coordinates of the last point and the (1+N1)th point on the main trajectory. V , satisfying |V V -V R |≤ΔVx, where ΔVx is the radial velocity threshold for trajectory matching, and its value ranges from [0,+∞).

[0041] Where α = α3 when mark_id = 1, and α = α1 when mark_id = 0, the speed calculation method is as follows: Iterate through each target trajectory to obtain a set of candidate trajectories for each target's main trajectory.

[0042] Preferably, step B2 further includes:

[0043] If the target trajectory matches multiple candidate trajectories, calculate the distance Δd between the trajectory points of the main trajectory and each trajectory point in the candidate trajectories within the marking period, and sum the Δd for each trajectory point to obtain Δd. sum By traversing each candidate trajectory, multiple Δd values ​​are obtained. sum Δd sum The smallest candidate trajectory is the trajectory that uniquely connects to the main trajectory of the target; if Δd sum If there are multiple candidate trajectories with the smallest Δd, then the candidate trajectory with the smallest Δd value is considered the trajectory uniquely connected to the target main trajectory. This process is repeated for each target trajectory to obtain a unique candidate trajectory for each target trajectory. If a candidate trajectory meets the connection conditions of multiple target main trajectories, it is then compared with Δd... sum Minimal target trajectory connection.

[0044] Where Δd = ΔLon + ΔLat.

[0045] Preferably, the addition of virtual points further includes:

[0046] Calculate the target trajectory velocity and position information: Obtain the latitude, longitude, and coordinates of the nearest trajectory point and the (1+N4)th trajectory point on the main trajectory, and calculate the longitude velocity V between the two trajectory points. lon Latitude speed V lat K is the ratio of the difference in x-distance to the difference in y-distance between two trajectory points;

[0047] Virtual points are added based on information: If |V lon |≤V0 or|V lat If |≤V0, where V0 is the simulated point velocity threshold, a configuration parameter with a value range of [0,+∞), and the cumulative number of virtual points on the main trajectory is less than N5, then one virtual point is added to the main trajectory; if |V lon |>V0 or|V lat |>V0, if the cumulative number of virtual points on the main trajectory is less than N5, and the connection fails after N6 attempts, add a virtual point to the main trajectory. If the cumulative number of virtual points on the main trajectory is greater than N5, delete the main trajectory from the target trajectory set. N5 and N6 are the candidate trajectory filtering thresholds 5 and 6, respectively, which are configuration parameters.

[0048] Virtual points refer to points manually added after a connection failure. One virtual point is added each cycle to ensure the integrity of subsequent trajectories; virtual points do not actually exist. The latitude, longitude, x, y coordinates, and detection time of the latest trajectory point n and the (1+N4)th trajectory point n4 of the main trajectory are extracted, respectively (lon... n ,lat n ,t n ) and (lonn4 ,lat n4 ,t n4 ), (x n ,y n ,t n ) and (x n4 ,y n4 ,t n4 ), calculate the latitude, longitude, velocity, and K of the two trajectory points:

[0049] Longitude speed: Latitude speed: If |V lon |>V0 or|V lat |>V0 and K<1: If the number of virtual points on the main trajectory is less than N5, and the connection fails after N6 attempts, add one virtual point (x) to the main trajectory. n ,y n ,t n +T); If the number of virtual points is greater than N5, then the main trajectory is deleted from the target trajectory set. If |V lon |>V0 or|V lat |>V0 and K>1: If the number of virtual points on the main trajectory is greater than N5, then delete the main trajectory from the target main trajectory set; if the number of virtual points on the main trajectory is less than N5, add virtual points (lon) to the main trajectory. n +T*V lon ,lat n +T*V lat ,t n +T).

[0050] Therefore, the present invention has the following beneficial effects: by deploying radars in the middle section of the road and connecting the radar trajectory with the radar trajectory and connecting multiple radar trajectories, complete trajectory information of vehicles in the middle section of the road is obtained. The accuracy is high, the matching is fast, and the jump is effectively prevented. At the same time, it can provide an important reference for the detection and management of urban road traffic safety. Attached Figure Description

[0051] Figure 1 This is a flowchart of the steps of the method of the present invention.

[0052] Figure 2 This is a flowchart illustrating the specific operation of radar deployment and area delineation in the method of the present invention.

[0053] Figure 3 This is a parameter value table for Embodiment 1 of the method of the present invention.

[0054] Figure 4 This is a flowchart illustrating the overall implementation of the trajectory connection method of the present invention.

[0055] Figure 5This is a schematic diagram illustrating the connection between the method of the present invention and the radar trajectory;

[0056] Figure 6 This is a schematic diagram of the multi-radar trajectory connection method of the present invention. Detailed Implementation

[0057] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0058] Example 1:

[0059] like Figure 1 The illustrated embodiment demonstrates a method for optimizing trajectory connection across multiple radar detection areas on a road segment. The operation process is as follows: Step 1, selecting radar deployment points and defining the radar connection area; Step 2, acquiring the set of target trajectories detected by the radar, filtering the target trajectory set to obtain a deduplicated set of target trajectories; Step 3, determining whether the deduplicated set of target trajectories falls within a multi-radar connection area and whether trajectories have abruptly changed, and performing same-radar trajectory connection or multi-radar trajectory connection based on the determination results; Step 4, traversing all target trajectories, analyzing the remaining candidate trajectories, and obtaining the complete trajectory set of each target vehicle. By deploying radars in the middle road segment for same-radar trajectory connection and multi-radar trajectory connection, complete trajectory information of vehicles in the middle road segment is obtained, achieving high accuracy, fast matching, and effectively preventing abrupt changes. This method can also provide important reference for the detection and management of urban road traffic safety.

[0060] The specific embodiments of the present invention will be further described in detail below through concrete examples:

[0061] Select radar deployment locations and define the radar connection area.

[0062] Based on the radar detection capabilities, radar deployment points are selected on the road segment. Two radars should be deployed at the same radar deployment point to detect two opposing traffic flows. For two adjacent radar deployment points, the area in between is the multi-radar area, and the detection ranges of the two radars in the multi-radar area need to overlap. The deployment is completed for the entire road segment to achieve full coverage of the road segment detection range.

[0063] If radar X1 and radar X2 are millimeter-wave radars installed at two adjacent radar deployment points, then the cross detection range of radar X1 and radar X2 is defined as the multi-radar connection area Z0, the area where radar X1 does not cross with radar X2 is defined as the same-radar connection area Z1, and the area where radar X2 does not cross with radar X1 is defined as the same-radar connection area Z2.

[0064] Taking two intersections in City A as an example, radar X1 and radar X2 are millimeter-wave radars deployed at adjacent radar deployment points in a multi-radar area between the two intersections; for example... Figure 2As shown, the cross detection range of radar X1 and radar X2 in the middle section is defined as the multi-radar connection area Z0. The area in the middle section where the detection range of radar X1 does not intersect with radar X2 is defined as the same radar connection area Z1. The area in the middle section where the detection range of radar X2 does not intersect with radar X1 is defined as the same radar connection area Z2.

[0065] All parameter values ​​used in this embodiment are as follows: Figure 3 As shown.

[0066] The method for connecting the regional trajectories of two vehicles traveling in different directions within a multi-radar area is the same. To achieve real-time trajectory connection, let the current time be t and the calculation period be T. Therefore, the current period is defined as (tT, t], and the previous period is defined as (t-2*T, tT]. Taking the travel directions from radar X1 and radar X2 as an example:

[0067] According to the trajectory connection method proposed in this invention, road segment trajectory connection is performed. For each target vehicle trajectory, the connection process is as follows: Figure 4 As shown: Filter the target area and traverse all target trajectories: Determine whether the area where the target vehicle is located is a multi-radar connection area. If so, connect it with the radar trajectory. Then filter candidate trajectories and connect multi-radar trajectories. If not, determine whether a trajectory jump has occurred. If so, connect it with the radar trajectory. Until all target trajectories have been traversed, analyze the remaining candidate trajectories to obtain the complete trajectory of the road segment.

[0068] The specific steps are as follows:

[0069] 1. Obtain the set of different target trajectories in the current period: Obtain the set of target trajectories detected by the radar, filter the set of target trajectories, and obtain the set of target trajectories after deduplication.

[0070] (1) Extract the set of radar vehicle IDs (vehicle_id) from the radar X1 detection data that satisfy the condition that the radar trajectory time (detect_time) belongs to (t-2*T, tT], and the latest trajectory latitude and longitude are within Z1. Filter to obtain the target trajectory set located in the multi-radar connection area Z0 at the current time. After deduplication, it is denoted as set A. Filter to obtain the target trajectory set located in the same radar connection area Z1 at the current time. After deduplication, it is denoted as set B.

[0071] (2) Extract the set of radar vehicle numbers from the radar X2 detection data that satisfy the condition that the radar trajectory time belongs to (t-2*T, tT], and the latest point trajectory latitude and longitude are within Z2. After filtering and deduplicating the target trajectory set located in the same radar connection area Z2 at the current time, denote it as set C.

[0072] Second, determine whether the deduplicated target trajectory set is a multi-radar connection area and whether the trajectory has changed. Based on the judgment result, connect the same radar trajectory or the multi-radar trajectory.

[0073] (1) Multi-radar connection area: Connect the trajectories of each target in set A:

[0074] 1) Connect with radar trajectory: Obtain the new trajectory points of each target trajectory within the time period (tT,t], traverse each target trajectory in set A, and store each target trajectory and its detection status information in set A;

[0075] 2) Obtain the set of target trajectories that satisfy the following conditions: the total number of trajectory points in the time period (tT,t] is greater than or equal to N1, the target trajectory has trajectory points and belongs to the multi-radar connection area Z0, and the X coordinate value of the latest trajectory point is less than the X coordinate value of the 1+N4th trajectory point (sorted in ascending order of time). After deduplication, denote it as D.

[0076] 3) Perform multi-radar trajectory integration and save the successfully integrated target trajectories in set G.

[0077] (2) In non-multi-radar connection areas, the status of each target trajectory in set B and set C is judged. If there are no newly generated trajectory points in (tT,t], the main trajectory is judged to have changed, and the connection with the radar trajectory is performed. The successfully connected target trajectory is saved in set G. If there are newly generated trajectory points in (tT,t], the main trajectory has not changed. Set B is traversed, and the latitude, longitude, X coordinate (x_pos), Y coordinate (y_pos), and timestamp of each target trajectory in (tT,t] are saved in set G.

[0078] The specific implementation method for trajectory connection is as follows:

[0079] 1. Implementation method for connecting with radar trajectory.

[0080] (1) Filter the candidate trajectory set of the target trajectory based on the spatiotemporal range.

[0081] For the same connecting target, the main trajectory is denoted as L0, and the candidate trajectory is denoted as L1, such as... Figure 5 As shown, vehicle_id and its detection status information that meet the following conditions are obtained from the detection data:

[0082] a. The candidate trajectory L1 trajectory points and the trajectory points before the jump of the target main trajectory L0 are detected by the same radar;

[0083] b. The first trajectory point of candidate trajectory L1 is later than the last trajectory point of target main trajectory L0, and the time difference is within the agreed range, i.e., 0. <t1-t0<T1;

[0084] c. The difference in longitude and latitude between the first trajectory point of candidate trajectory L1 and the last trajectory point of target main trajectory L0 is within the range of ΔLon1 and ΔLat1.

[0085] d. The number of trajectory points of candidate trajectory L1 in (tT,t] is greater than or equal to N3.

[0086] Based on the above four conditions, a set of candidate trajectories for the target trajectory is obtained. If the number of candidate trajectories is 0, the matching fails and proceeds to (4); otherwise, proceeds to (2).

[0087] (2) Based on trajectory features, the candidate trajectory set of the target trajectory obtained in (1) is further filtered.

[0088] Candidate trajectories must also meet the following conditions:

[0089] (1) Trajectory direction angle matching: It should be ensured that the vector formed by the first point and the first + N4th point on the candidate trajectory and the vector formed by the last point and the last + N4th point on the target main trajectory have an angle less than or equal to α1.

[0090] (2) Connection direction angle matching: It should be ensured that the vector formed by the last point on the target trajectory and the first point on the candidate vehicle trajectory, and the vector formed by the last point on the target vehicle trajectory and the last + N4 points from the end, have an angle less than α2.

[0091] Obtain the set of all candidate trajectories that meet the conditions. If the number of candidate trajectories is 0, proceed to (4); otherwise, proceed to (3).

[0092] (3) Connect candidate trajectories.

[0093] 1) Filter the target trajectory to find the only candidate trajectory.

[0094] Among all candidate trajectories that meet the spatiotemporal range and trajectory characteristics, the candidate target whose first trajectory point is closest to the last trajectory point of the target main trajectory is selected, i.e., the smallest Δd = ΔLon + ΔLat. The candidate trajectory with the smallest value is matched with the target main trajectory, the ID of the target main trajectory is assigned to the candidate trajectory, and the connection between the target main trajectory and the radar trajectory is ended.

[0095] 2) Filter candidate trajectories to find the unique target trajectory.

[0096] If a candidate trajectory meets the connection conditions of multiple target trajectories, it is connected with the target trajectory with the smallest Δd; if the connection is successful, a label mark_id is added to the target trajectory and mark_id=1, and the connection with the radar trajectory is completed; if the connection fails, proceed to (4).

[0097] (4) Handling connection failures.

[0098] 1) If the latest point of the main trajectory is located in the multi-radar connection area Z0, add a label mark_id to the target trajectory and mark mark_id=0, and perform multi-radar area trajectory connection;

[0099] 2) If the latest point of the main trajectory is located in the same radar connection area Z1, then add a virtual point.

[0100] 2. Implementation method for multi-radar trajectory integration.

[0101] For a target trajectory with mark_id = 1, the marking period is (tT, t]; for a target trajectory with mark_id = 0, the marking period is (t-2*T, t+T]. A schematic diagram of the multi-radar trajectory integration implementation is shown below. Figure 6 As shown.

[0102] (1) Filter the candidate trajectory set of the target trajectory based on the spatiotemporal range.

[0103] For the same connecting target, the main trajectory is denoted as L2. The spatiotemporal parameters of the trajectory points within the marked period in set A are obtained. Spatiotemporal parameters of each point on the candidate trajectory Point matching is considered successful if the following conditions are met:

[0104] a. The time corresponding to the candidate trajectory points Inside, The time corresponding to the main trajectory point being matched (T2 = T2 when mark_id = 1, T2 = T2 + 1 when mark_id = 0);

[0105] b. The difference between the longitude and latitude of the candidate trajectory points and the longitude and latitude of the main trajectory points is within the range of ΔLon2 and ΔLat2 (ΔLat2 = ΔLat2 when mark_id = 1, ΔLat2 = ΔLat2 + 0.0001 when mark_id = 0).

[0106] (2) Based on trajectory features, the candidate trajectory set of the target trajectory obtained in (1) is further filtered.

[0107] The candidate trajectory set obtained in (1) for each target main trajectory is further filtered: take the spatiotemporal parameters of the latest point and the last + N4 points of the candidate trajectory, as well as the spatiotemporal parameters of the latest point and the last + N4 points of the main trajectory. If the main trajectory has two trajectory points within the marking period, take the spatiotemporal parameters of these two points. If the main trajectory has only one point within the marking period, take the spatiotemporal parameters of that point and the spatiotemporal parameters of the trajectory point with the longest time in the previous period of the target trajectory.

[0108] When mark_id = 1 or mark_id = 0, the direction angle constraint condition is satisfied: the vector formed by the first point and the first (1+N4)th point on the candidate trajectory and the vector formed by the last point and the last (1+N4)th point on the main trajectory have an angle less than or equal to α (α = α2 when mark_id = 1, α = α3 when mark_id = 0);

[0109] When mark_id = 1, the velocity constraint condition must also be met: calculate the velocity V using the X coordinates of the 1st point and the (1+N4th)th point on the candidate trajectory. R The speed calculation method is as follows: The velocity V is calculated using the X-coordinates of the last point and the (1+N1)th point on the main trajectory. V , satisfying |V V -V R |≤ΔVx;

[0110] Iterate through each target trajectory to obtain a set of candidate trajectories for each target's main trajectory.

[0111] (3) Connect candidate trajectories.

[0112] 1) Filter out candidate trajectories that have a unique target trajectory.

[0113] If the target trajectory matches multiple candidate trajectories, calculate the distance Δd between the trajectory points of the main trajectory and each trajectory point in the candidate trajectories within the marking period, where Δd = ΔLon + ΔLat. Then, sum the Δd values ​​for each trajectory point to obtain Δd. sum By traversing each candidate trajectory, multiple Δd values ​​are obtained. sum Δd sum The smallest candidate trajectory is the trajectory that uniquely connects to the main trajectory of the target; if Δd sum If there are multiple candidate trajectories with the smallest Δd, then the candidate trajectory with the smallest Δd value is the trajectory that is uniquely connected to the main trajectory of the target; traverse each target trajectory to obtain a unique candidate trajectory for each target trajectory.

[0114] 2) Filter the candidate trajectory to find the unique target trajectory.

[0115] If a candidate trajectory meets the connection conditions of multiple target trajectories, then it is connected with the smallest target trajectory.

[0116] (4) Expand the search scope to perform secondary matching.

[0117] If no suitable candidate trajectory is found, the search range is expanded appropriately, the coefficient ε is determined, and T′2=εT2, ΔLon′=εΔLon, ΔLat′=εΔLat, ΔV′ X =εΔVx, α3 remains unchanged, and perform secondary matching between the target trajectory and the candidate trajectory.

[0118] (5) Process the connection results.

[0119] 1) If a candidate trajectory is successfully matched, the spatiotemporal parameters of the latest trajectory point of the candidate trajectory are the spatiotemporal parameters of the latest trajectory point of the main trajectory within the marking period. The target ID of the main trajectory, the target ID of the connected candidate trajectory, and the detected state information are stored in set G.

[0120] 2) If no candidate trajectory is successfully matched, there are two possible scenarios:

[0121] For target trajectories with mark_id=1: the latest trajectory point obtained by connecting the original radar trajectory is retained, and the target ID of this type of target trajectory, the target ID of the candidate trajectory to be connected, and the detected state information are extracted and stored in set G;

[0122] For the target trajectory with mark_id=0: add a virtual point and store the target trajectory status information of the added virtual point into set G;

[0123] Remove candidate trajectories that have been successfully connected from the candidate trajectory set.

[0124] In this embodiment, virtual points are added in the following way:

[0125] (1) Calculate the target trajectory velocity and position information.

[0126] Extract the latitude, longitude, and coordinates of the latest trajectory point and the (1+N4)th trajectory point of the main trajectory, respectively (lon n ,lat n ,t n ) and (lon n4 ,lat n4 ,t n4 ), (x n ,y n ,t n ) and (x n4 ,y n4 ,t n4 ), calculate the latitude and longitude velocities of the two trajectory points:

[0127]

[0128] (2) Add virtual points based on information.

[0129] 1) If |V lon |≤V0 or|V lat If |≤V0, and the cumulative number of virtual points is less than N5, then the main trajectory adds one virtual point (x). n ,y n ,t n +T);

[0130] 2) If |V lon |>V0 or|V lat |>V0, K<1: If the number of virtual points on the main trajectory is less than N5, and the connection fails after N6 attempts, add one virtual point (x) to the main trajectory. n ,y n ,t n +T); If the number of virtual points is greater than N5, then the main trajectory will be removed from the target trajectory set;

[0131] 3) If |V lon |>V0 or|V lat |>V0, K>1: If the number of virtual points on the main trajectory is greater than N5, then delete the main trajectory from the target main trajectory set; if the number of virtual points on the main trajectory is less than N5, add virtual points (lon) to the main trajectory. n +T*V lon ,lat n +T*V lat ,t n +T).

[0132] 3. Traverse all target trajectories, analyze the remaining candidate trajectories, and obtain the complete trajectory of the road segment.

[0133] Delete the candidate trajectory data that has been successfully connected, store the detection status information of the remaining target trajectories with a number of trajectory points greater than or equal to N2 in set D in set D, and add them to set G if they are successfully connected; the set of target trajectories in set G is the complete trajectory set of each target vehicle in the area.

[0134] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Other variations and modifications are possible without departing from the technical solutions described in the claims.

Claims

1. A method for optimizing trajectory connection across multiple radar detection areas on a road segment, characterized in that, It includes the following steps: S1: Select radar deployment locations and define the radar connection area; S2: Obtain the set of target trajectories detected by the radar, filter the set of target trajectories, and obtain the set of target trajectories after deduplication; S3: Determine whether the deduplicated target trajectory set is a multi-radar connection area and whether the trajectory has changed. Based on the judgment result, connect the same radar trajectory or the multi-radar trajectory. When connecting multiple radar trajectories: B1: Based on the spatiotemporal range, a set of candidate trajectories for the main trajectory is obtained. The candidate trajectories for each target's main trajectory are further filtered based on trajectory features. Each target trajectory is traversed to obtain a set of candidate trajectories for each target's main trajectory. B2: Traverse each target trajectory to obtain a unique candidate trajectory for each target trajectory, and filter the candidate trajectory to find the unique main trajectory; B3: If no suitable candidate trajectory is found, expand the search range to perform a second matching between the target's main trajectory and the candidate trajectory; if no candidate trajectory is found, add virtual points to the target trajectory that fails to connect with the radar trajectory and whose latest point of the main trajectory is located in the multi-radar connection area Z0. If a candidate trajectory is successfully matched, the spatiotemporal parameters of the latest trajectory point of the candidate trajectory are the spatiotemporal parameters of the latest trajectory point of the main trajectory within the marking period. S4: Traverse all target trajectories, analyze the remaining candidate trajectories, and obtain the complete trajectory set of each target vehicle.

2. The method for optimizing trajectory connection in multiple radar detection areas of a road segment according to claim 1, characterized in that, Step S2 is further expressed as follows: S2.1: If radar X1 and radar X2 are millimeter-wave radars installed at two adjacent radar deployment points, obtain the target trajectory set detected by radar X1, filter the target trajectory set located in the multi-radar connection area Z0 at the current time, and deduplicate it as set A; The set of target trajectories located in the same radar interface region Z1 at the current moment is filtered, and after deduplication, it is denoted as set B; S2.2: Obtain the set of target trajectories detected by radar X2, filter the set of target trajectories located in the same radar junction area Z2 at the current time, and deduplicate them, denoted as set C.

3. The method for optimizing trajectory connection in multiple radar detection areas of a road segment according to claim 2, characterized in that, Step S3 is further expressed as follows: For the multi-radar connection area: perform trajectory connection for each target trajectory in set A: perform trajectory connection with the same radar trajectory to obtain new trajectory points for each target trajectory within the time period (tT, t], and update each target trajectory in set A; filter and deduplicate the candidate trajectory set detected by radar X2: for each target trajectory, it must satisfy that the total number of trajectory points within the time period (tT, t] is greater than or equal to N1 and there are trajectory points belonging to the multi-radar connection area Z0, to obtain set D; Perform multi-radar trajectory integration, save the successfully integrated target trajectories in set G, N1 is the candidate trajectory screening threshold 1, t is the current time, and T is the calculation period; Not a multi-radar connection area: Perform state judgment on each target trajectory in set B and set C: If the target trajectory does not generate a new trajectory point within the time period (tT, t], then the target trajectory changes and is connected with the radar trajectory. The successfully connected target trajectory is saved in set G. Conversely, if the target trajectory does not change, the detection state information corresponding to the target trajectory within (tT, t] is stored in set G.

4. A method for optimizing trajectory connection in multiple radar detection areas along a road segment according to claim 1, 2, or 3, characterized in that, In step S3, the connection with the radar trajectory further includes: A1: Based on the spatiotemporal range, a set of candidate trajectories for the main trajectory is obtained. Each target main trajectory is further filtered in the set of candidate trajectories based on trajectory features. A2: Calculate the distance ∆d between the first trajectory point of each target on the candidate trajectory and the last trajectory point of the main trajectory. Select the candidate trajectory with the smallest ∆d and match it with the main trajectory. Assign the ID of the main trajectory to the candidate trajectory and end the connection between the main trajectory and the radar trajectory of the target. If a candidate trajectory meets the connection conditions of multiple target trajectories, connect it with the target trajectory with the smallest ∆d. If the connection is successful, add a label mark_id to the target trajectory and mark mark_id = 1. The connection with the radar trajectory is completed. If the connection fails, proceed to A3. A3: If the latest point of the main trajectory is located in the multi-radar connection area Z0, add a label mark_id to the target trajectory and mark mark_id = 0, and perform multi-radar area trajectory connection; if the latest point of the main trajectory is located in the same radar connection area Z1, add a virtual point.

5. The method for optimizing trajectory connection in multiple radar detection areas of a road segment according to claim 4, characterized in that, Step A1 is further expressed as follows: A1.1: Based on the spatiotemporal range conditions: the trajectory points of candidate trajectory L1 and the trajectory points of the main trajectory L0 before the jump are detected by the same radar; the time of the first trajectory point of candidate trajectory L1 is later than the time of the last trajectory point of the main trajectory L0, and the time difference is within the range of (0, T1), where T1 is the trajectory matching time threshold 1; the difference in longitude and latitude between the first trajectory point of candidate trajectory L1 and the last trajectory point of the main trajectory L0 is within the range of ∆Lon1 and ∆Lat1. The number of trajectory points of candidate trajectory L1 in (tT, t] is greater than or equal to N3, where N3 is the candidate trajectory filtering threshold of 3; obtain the candidate trajectory set. If there are 0 candidate trajectories, the matching fails and proceed to A3; otherwise, proceed to A1.

2. A1.2: Select a set of candidate trajectories that satisfy the following conditions: the angle between the vector formed by the first point and the (1+N4)th point on the candidate trajectory and the vector formed by the last point and the (1+N4)th point from the end on the main trajectory is less than or equal to α1, where α1 is the trajectory matching angle threshold 1; and the angle between the vector formed by the last point on the main trajectory and the first point on the candidate vehicle trajectory and the vector formed by the last point and the (1+N4)th point from the end on the main trajectory is less than α2, where α2 is the trajectory matching angle threshold 2. If the candidate trajectory is 0, the matching fails and proceeds to A3; otherwise, proceeds to A2. N4 is the candidate trajectory filtering threshold 4.

6. The method for optimizing trajectory connection in multiple radar detection areas of a road segment according to claim 1, characterized in that, The multi-radar trajectory connection further includes storing the target ID of the main trajectory, the target ID of the candidate trajectory to be connected, and the detected state information into a set G.

7. The method for optimizing trajectory connection in multiple radar detection areas of a road segment according to claim 6, characterized in that, Step B1 further includes: Based on spatiotemporal range: For the same connecting target, obtain the spatiotemporal parameters of the trajectory points within the marked period in the target trajectory set A of the deduplicated multi-radar connecting area Z0 of the main trajectory. Traverse the spatiotemporal parameters of each point in the candidate trajectory to perform point matching. If the time corresponding to the candidate trajectory point is within [t... iR_1 -T i , t iR_1 +T i Within ], or the difference between the longitude and latitude of the candidate trajectory point and the longitude and latitude of the main trajectory point within ∆Lon i With ∆Lat i If the location is within the specified range, it indicates a successful point match; t iR_1 T represents the current detection time of the candidate trajectory point. i To match the configuration parameters, ∆Lon i To match the radar matching parameter distance threshold, ∆Lat i For multi-radar matching parameters, the distance threshold is used; Based on trajectory features: take the spatiotemporal parameters of the latest point and the 1+N4th point from the end of the candidate trajectory, as well as the spatiotemporal parameters of the latest point and the 1+N4th point from the end of the main trajectory. If the main trajectory has two trajectory points within the marking period, take the spatiotemporal parameters of these two points; if the main trajectory has only one point within the marking period, take the spatiotemporal parameters of that point and the spatiotemporal parameters of the trajectory point with the longest time in the previous period of this marking period of the target trajectory.

8. A method for optimizing trajectory connection in multiple radar detection areas along a road segment according to claim 6 or 7, characterized in that, When filtering based on trajectory features, constraints are also included: For target trajectories that are successfully matched with radar trajectories or fail to be matched with radar trajectories but whose latest point is located in the multi-radar connection area Z0, the following azimuth constraint condition is satisfied: the angle between the vector formed by the first point and the (1+N4)th point on the candidate trajectory and the vector formed by the last point and the (1+N4)th point from the end on the main trajectory is less than or equal to α. For a target trajectory that successfully matches the radar trajectory, the velocity constraint condition is satisfied: the velocity V is calculated using the X coordinates of the 1st point and the (1+N4th)th point on the candidate trajectory. R The velocity V is calculated using the X-coordinates of the last point and the (1+N1)th point on the main trajectory. V , satisfying |V V -V R |≤∆Vx, where ∆Vx is the radial velocity threshold for trajectory matching.

9. A method for optimizing trajectory connection in multiple radar detection areas along a road segment according to claim 6 or 7, characterized in that, Step B2 further includes: If the target trajectory matches multiple candidate trajectories, calculate the distance ∆d between the trajectory points of the main trajectory and each trajectory point in the candidate trajectories within the marking period, and sum the ∆d of each trajectory point to obtain ∆d. sum Iterate through each candidate trajectory to obtain multiple ∆d. sum ,∆d sum The smallest candidate trajectory is the trajectory that uniquely connects to the target trajectory; if ∆d sum If there are multiple candidate trajectories with the minimum ∆d, then the candidate trajectory with the minimum ∆d is considered the unique connecting trajectory to the target trajectory. This process is repeated for each target trajectory to obtain a unique candidate trajectory for each target trajectory. If a candidate trajectory meets the connecting conditions of multiple target trajectories, it is then compared with ∆d. sum Minimal target trajectory connection.

10. The method for optimizing trajectory connection in multiple radar detection areas of a road segment according to claim 4, characterized in that, The addition of virtual points further includes: Calculate the target trajectory velocity and position information: Obtain the latitude, longitude, and coordinates of the nearest trajectory point and the (1+N4)th trajectory point on the main trajectory, and calculate the longitude velocity V between the two trajectory points. lon Latitude speed V lat K is the ratio of the difference in x-distance to the difference in y-distance between two trajectory points; Virtual points are added based on information: If |V lon |≤V0 or |V lat If |≤V0, where V0 is the simulated point velocity threshold, and the cumulative number of virtual points on the main trajectory is less than N5, then one virtual point is added to the main trajectory; if |V lon |>V0 or |V lat If the cumulative number of virtual points on the main trajectory is less than N5 and the connection fails after N6 attempts, add a virtual point to the main trajectory. If the cumulative number of virtual points on the main trajectory is greater than N5, delete the main trajectory from the target trajectory set. N5 is the candidate trajectory selection threshold of 5, and N6 is the candidate trajectory selection threshold of 6.

Images