Target tracking method, device, equipment, medium and product

Abstract

The invention discloses a target tracking method, device and equipment, a medium and a product. The method comprises the following steps: receiving point cloud data sent by roadside equipment; identifying the point cloud data to obtain a plurality of targets and traffic states; if the traffic state is a congestion state and there is a lost target, generating a virtual target corresponding to the lost target; target tracking is carried out based on the virtual target corresponding to the lost target and the multiple targets, and through the technical scheme of the invention, the accuracy of target tracking can be improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a target tracking method, apparatus, device, medium and product. Background Technology

[0002] In intelligent traffic monitoring systems, millimeter-wave radar has become one of the core sensors for roadside and vehicle-side perception due to its advantages such as accurate ranging and speed measurement and immunity to the effects of lighting and weather. However, existing millimeter-wave radar-based perception algorithms typically employ preset fixed parameters and strategies for target clustering and tracking. While these methods work effectively in scenarios with smooth traffic flow and sparse targets, their accuracy decreases during traffic congestion. Summary of the Invention

[0003] This invention provides a target tracking method, apparatus, device, medium, and product that can improve the accuracy of target tracking.

[0004] According to one aspect of the present invention, a target tracking method is provided, comprising:

[0005] Receive point cloud data sent by roadside equipment;

[0006] The point cloud data is then identified to obtain multiple targets and traffic conditions;

[0007] If the traffic is congested and there is a lost target, then a virtual target corresponding to the lost target is generated.

[0008] Target tracking is performed based on the virtual target corresponding to the lost target and multiple targets.

[0009] According to another aspect of the present invention, a target tracking device is provided, the target tracking device comprising:

[0010] The point cloud data receiving module is used to receive point cloud data sent by roadside equipment;

[0011] The point cloud data recognition module is used to recognize the point cloud data to obtain multiple targets and traffic conditions;

[0012] The virtual target generation module is used to generate virtual targets corresponding to lost targets if the traffic status is congested and lost targets exist.

[0013] The target tracking module is used to track targets based on the virtual target corresponding to the lost target and multiple targets.

[0014] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0015] At least one processor; and

[0016] A memory communicatively connected to the at least one processor; wherein,

[0017] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the target tracking method according to any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the target tracking method according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer program product is provided, which, when executed by a processor, implements the target tracking method as described in any of the embodiments of the present invention.

[0020] This invention receives point cloud data sent by roadside equipment; identifies the point cloud data to obtain multiple targets and traffic conditions; if the traffic condition is congested and there are lost targets, a virtual target corresponding to the lost target is generated; target tracking is performed based on the virtual target corresponding to the lost target and multiple targets, which can solve the problem of losing stationary or low-speed targets when the traffic condition is congested. When the traffic condition is congested, generating a virtual target of the lost target and performing target tracking based on the virtual target and multiple targets can improve the accuracy of target tracking.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a target tracking method according to an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the structure of a target tracking device according to an embodiment of the present invention;

[0025] Figure 3This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0029] Example 1

[0030] Figure 1 This is a flowchart illustrating a target tracking method provided in an embodiment of the present invention. This embodiment is applicable to target tracking situations. The method can be executed by the target tracking device in this embodiment of the invention, which can be implemented in software and / or hardware, such as... Figure 1 As shown, the method specifically includes the following steps:

[0031] S110 receives point cloud data sent by roadside equipment.

[0032] In this embodiment, the point cloud data can be point cloud data collected by millimeter-wave radar in roadside equipment.

[0033] S120, the point cloud data is identified to obtain multiple targets and traffic conditions.

[0034] In this embodiment, the method for identifying multiple targets and traffic conditions from the point cloud data can be as follows: target detection is performed on the point cloud data to obtain multiple targets; lane line identification is performed based on the point cloud data to obtain the number, position, and width of lane lines; target density in the current lane is calculated; the average speed of all targets in the lane is calculated; and traffic conditions are determined based on the target density in the current lane and the average speed of all targets in the lane.

[0035] In this embodiment, lane line recognition based on the point cloud data can be performed as follows: Self-identification of lane lines based on radar's own perception information: Based on historical multi-frame target track point information, a lane line calibration method based on Hough transform is used to identify the number, position, and width of lane lines. Manual prior information: Manually drawn lane line information is input and converted into a self-identified lane line structure. Fusion and correction mechanism: The self-identification result is compared and fused with the manual prior information. If the self-identification result is consistent with the manual information within the tolerance range, the lane line confidence is enhanced. If the self-identified lane lines have anomalies such as intersections, position / width, or significant differences from the manual lane line parameters (slope, width, length, position, etc.), the parameters of the manual lane lines are used to correct the self-identified lane lines, ensuring the real-time performance and accuracy of the lane line information.

[0036] It should be noted that by fusing and correcting manual lane lines with self-recognition lane line algorithms, the accuracy and real-time performance of lane line self-recognition can be improved.

[0037] In this embodiment, the traffic status is determined by identifying the point cloud data as follows: Statistical target density D: Calculate the number of vehicles per unit length (e.g., per kilometer) within the current lane. Statistical average speed V: Calculate the average speed of all targets within the lane. Low speed is a core characteristic of congestion. Output traffic status label: Based on the target density and average speed characteristics within the lane, a threshold-based logical judgment method is used to identify congestion scenarios and output traffic status labels (smooth / congested). That is, when the target density D > density threshold Dt and the average speed V > speed threshold Vt, it is determined to be a congested scenario.

[0038] S130, if the traffic is congested and there is a lost target, then generate a virtual target corresponding to the lost target.

[0039] In this embodiment, if the traffic is congested, it is determined whether a lost target exists. If a lost target exists, a virtual target corresponding to the lost target is generated. It should be noted that the method for determining whether a lost target exists can be: determining whether a lost target exists based on two adjacent frames of point cloud data. For example, if target A exists in the previous frame of point cloud data, but target A does not exist in the current frame of point cloud data, then target A is determined to be lost. Another method for determining whether a lost target exists can be: if there are two targets in a stop-and-go state within the same lane, and the spatial distance between the two targets in a stop-and-go state is greater than a fourth distance threshold and less than a fifth distance threshold, where the fourth distance threshold is less than the fifth distance threshold; or, if there are no other targets in front of the targets in a stop-and-go state, and the distance between the targets in a stop-and-go state and the stop line is greater than a sixth distance threshold, then a lost target is determined to exist.

[0040] In this embodiment, if the traffic is congested, and a lost target is determined based on two adjacent frames of point cloud data, and there are no other targets in front of the lost target but a target behind it, then a virtual target corresponding to the lost target is generated based on the movement speed of the target behind the lost target. If the traffic is congested, and a lost target is determined based on two adjacent frames of point cloud data, and there are no other targets in front of or behind the lost target, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target. If the traffic is congested, and a lost target is determined based on two adjacent frames of point cloud data, and the target in front of the lost target is in a first state, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target. If the traffic is congested, and a lost target is determined based on two adjacent frames of point cloud data, and the target in front of the lost target is in a second state, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target and the movement speed of the target in front of the lost target.

[0041] In this embodiment, after generating the virtual target corresponding to the lost target, it is necessary to synchronously update the motion attributes of the virtual target and add a maintenance flag to the newly generated virtual target.

[0042] It should be noted that when traffic is congested, extrapolated speed is one of the key factors in forcibly maintaining slow-moving targets with lost detection points. Therefore, if traffic is congested and lost targets exist, a virtual target corresponding to the lost target is generated based on at least one of the following: the speed of targets behind the lost target, the speed of targets in front of the lost target, and the historical speed of the lost target.

[0043] In this embodiment, after generating virtual targets, an independent track is started for each virtual target, a unique temporary ID is assigned, and a tracking filter is initialized. The track of the virtual target is then added to the track list. It should be noted that the track list includes: moving tracks, stop-and-go tracks, stationary tracks, etc.

[0044] Optionally, if the traffic situation is congested and there are lost targets, then a virtual target corresponding to the lost target is generated, including:

[0045] If the traffic is congested, there is a lost target, and there are no other targets in front of the lost target but there are targets behind it, then a virtual target corresponding to the lost target is generated based on the movement speed of the targets behind the lost target. This continues until the distance between the generated virtual target corresponding to the lost target and the stop line is less than the first distance threshold. Then, the movement attribute of the virtual target is changed from a moving state to a moving-stopping state, and a maintenance flag is added.

[0046] In this embodiment, if the lost target is the first target, there are no other targets in front, but there are targets behind, for stable longitudinal targets, the movement speed of the targets behind the lost target is forcibly extrapolated until it is extrapolated to the vicinity of the stop line. The movement attribute of the virtual target is forcibly changed from the movement state to the walking-stop state, and a forced maintenance mark is assigned.

[0047] It should be noted that the motion attribute of virtual targets not near the stop line is in motion state, while the motion attribute of virtual targets near the stop line is forcibly changed from motion state to moving-stop state.

[0048] In this embodiment, the distance between the virtual target and the stop line is less than the first distance threshold, indicating that the virtual target is near the stop line.

[0049] In this embodiment, the stop line can be a pedestrian crossing. The stop line identification method can be as follows: Statistical target speed characteristics: Statistically analyze the average speed and speed distribution of targets within the lane. In highway scenarios, vehicle speeds are high and variances are small; in urban roads, speeds are lower and stop-and-go traffic is frequent. Statistical target movement direction: Analyze the target's heading angle. At intersections, the movement directions of targets in different lanes differ significantly (straight, left turn, right turn). Output road scene labels: Based on the target speed and movement direction within the lane, output the road type, such as: highway, urban arterial road, intersection, etc. Special marker recognition: By associating with the target's historical trajectory, statistically analyze the distance D1 where multiple targets stop at the same location, the starting distance D2 of the longitudinal lane line, and the farthest projection point D3 of the trajectory of vehicles traveling laterally at the intersection in the longitudinal direction. Weighted fusion of these three statistical distances identifies the final stop line position Lstop, i.e., Lstop = w1 × D1 + w2 × D2 + w3 × D3, where w1, w2, and w3 are weights.

[0050] If the traffic is congested, there is a lost target, and there are no other targets in front of or behind the lost target, then a virtual target corresponding to the lost target is generated based on the lost target's historical movement speed. This continues until the distance between the generated virtual target and the stop line is less than a first distance threshold. Then, the motion attribute of the virtual target is changed from a moving state to a moving-stopping state, and a maintenance flag is added.

[0051] In this embodiment, the first distance threshold can be a preset distance, and this embodiment of the invention does not impose any restrictions on it.

[0052] In this embodiment, if the lost target is the first target and there are no other targets in front or behind, for stable longitudinal targets, the historical movement speed of the lost target is used as the basis. Force extrapolation continues until the virtual target is near the stop line. The motion attribute of the virtual target is forcibly changed from a moving state to a walking-stopping state, and a forced maintenance mark is assigned.

[0053] If the traffic is congested and there is a lost target, and the target in front of the lost target is in the first state, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target. This continues until the distance between the generated virtual target corresponding to the lost target and the target in front of the lost target is less than a second distance threshold. Then, the motion attribute of the virtual target is changed from the moving state to the walking-stopping state, and a maintenance flag is added.

[0054] In this embodiment, the first state can be a walking-stopping state or a stationary state. The second distance threshold can be a preset distance, for example, the second distance threshold can be 10 meters.

[0055] In this embodiment, if the lost target is not the first target, and the target ahead is in a moving / stopped state, then the lost target's historical speed is used as the determining factor. Forcefully push outward, slowly move to a distance of about N meters (e.g., 10 meters) from the target in front and stop. The virtual target's motion attribute changes from moving state to walking-stopping state, and a forced maintenance mark is assigned.

[0056] If the traffic is congested, there is a lost target, and the target in front of the lost target is in the second state, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target and the movement speed of the target in front of the lost target.

[0057] In this embodiment, the second state is a slow-moving state.

[0058] In this embodiment, the method for generating a virtual target corresponding to the lost target based on the historical movement speed of the lost target and the movement speed of the target in front of the lost target can be as follows: the historical movement speed of the lost target and the movement speed of the target in front of the lost target are weighted and summed to obtain the target movement speed; and the virtual target corresponding to the lost target is generated based on the target movement speed.

[0059] In this embodiment, when a target is determined to be lost, the lost target is not deleted immediately. Instead, a "congestion maintenance mode" is activated. The target's motion state (position, speed, acceleration, heading angle, etc.), key lane information, and the indexes of vehicles in front and behind are memorized before the target is lost. Based on the memorized information and a vehicle behavior prediction model in a congestion scenario (such as a following model), the target trajectory is continuously extrapolated (Coasting) through a filter and reported as a valid target, thereby solving the problem of lost stationary / low-speed targets.

[0060] Optionally, a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target and the movement speed of targets in front of the lost target, including:

[0061] The target's speed is obtained by weighted summing the historical speed of the lost target and the speed of the target ahead of the lost target.

[0062] A virtual target corresponding to the lost target is generated based on the target's movement speed.

[0063] In this embodiment, if the lost target is not the first target, and the target ahead is in a slow-moving state, then according to... - The filtered velocity is extrapolated. If the state of the target ahead of the lost target changes to a stop-and-go state, the target slowly moves to a distance of about 10m from the target ahead of the lost target and stops. The virtual target's motion attribute changes from a moving state to a stop-and-go state, and a forced maintenance flag is assigned. - Filtering speed is .in, , All are weights.

[0064] It should be noted that if the traffic is congested and a target in a stop-and-go state meets the deletion criteria, then the virtual target in the stop-and-go state is deleted. The deletion criteria include: the relative distance between the virtual target's track and each moving track in the track list is less than a sixth distance threshold. If the traffic is not congested, then targets that have continuously lost M frames and are in a stop-and-go state are deleted.

[0065] For example, if the traffic is congested, the virtual target is deleted if it meets the deletion criteria; otherwise, it is forcibly maintained. Deletion criteria: Iterate through the track list and calculate the speed V of each moving track. O >Speed ​​threshold V T The relative distance between the moving track and the target's track (Distance). If the moving track is close to the target's track (Distance < distance threshold D), then... T If the target loses M frames consecutively while moving or stopping in a non-congested scenario, deletion will be triggered.

[0066] Optionally, if the traffic situation is congested and a lost target exists, before generating the virtual target corresponding to the lost target, the following steps are also included:

[0067] If there are two targets in a stop-and-go state within the same lane, and the spatial distance between the two targets is greater than a third distance threshold and less than a fourth distance threshold, wherein the third distance threshold is less than the fourth distance threshold,

[0068] Alternatively, if there are no other targets in front of the target in a stopped state, and the distance between the target in a stopped state and the stop line is greater than the fifth distance threshold,

[0069] This confirms that a target has been lost.

[0070] In this embodiment, the third distance threshold, the fourth distance threshold, and the fifth distance threshold are all preset distances. The third distance threshold is smaller than the fourth distance threshold. For example, the third distance threshold can be 15 meters, the fourth distance threshold can be 50 meters, and the fifth distance threshold can be 40 meters.

[0071] In a specific example, if two stable, stop-and-go targets (Car_A and Car_B, assuming Car_A is in front of Car_B) are identified in the same lane, and the spatial distance Dis between the two targets is greater than the first threshold T1 (e.g., T1 = 15 meters) and less than the second threshold T2 (e.g., T2 = 50 meters), then it is determined that there is a lost target between the two stop-and-go targets.

[0072] In another specific example, if a stable target (lead vehicle) in a stop-and-go state is identified, with no other targets in front of it, and the distance Dis of the stop-and-go target from the stop line is greater than a threshold T (e.g., T = 40 meters), then it is determined that there is a lost target between the stop-and-go target and the stop line.

[0073] It should be noted that if the aforementioned situation of lost targets exists, target interpolation will be performed. Target interpolation is a supplement to target maintenance. In congested conditions, if the distance difference between two stable stop-and-go targets in the same lane is large, or if a stable stop-and-go target is far from the stop line, target interpolation will be performed. Here, a maximum allowable number of interpolated targets needs to be set to control risk.

[0074] Optionally, if the traffic situation is congested and there are lost targets, then a virtual target corresponding to the lost target is generated, including:

[0075] If the traffic is congested and there are lost targets, the interpolation quantity is determined based on the distance between two targets in a stop-and-go state, and a virtual target corresponding to the lost target is generated based on the interpolation quantity.

[0076] or,

[0077] If the traffic is congested and there are lost targets, the number of interpolations is determined based on the distance between the target in the stop-and-go state and the stop line, and a virtual target corresponding to the lost target is generated based on the number of interpolations.

[0078] In this embodiment, if there are two targets in a stop-and-go state within the same lane, and the spatial distance between the two targets is greater than a third distance threshold but less than a fourth distance threshold, then the interpolation quantity is determined based on the spatial distance between the two targets, and a virtual target corresponding to the lost target is generated based on the interpolation quantity. If there are no other targets in front of the target in a stop-and-go state, and the distance between the target and the stop line is greater than a fifth distance threshold, then the interpolation quantity is determined based on the distance between the target and the stop line, and a virtual target corresponding to the lost target is generated based on the interpolation quantity.

[0079] In this embodiment, the method for determining the interpolation quantity based on the distance between two targets in a stop-and-go state can be as follows: determine the interpolation quantity based on the distance between the two targets in a stop-and-go state and the expected average vehicle distance.

[0080] In this embodiment, the method for determining the interpolation quantity based on the distance between the target in a stop-and-go state and the stop line can be as follows: determine the interpolation quantity based on the distance between the target in a stop-and-go state and the stop line and the expected average vehicle distance.

[0081] In a specific example, the initial number of interpolations N_calculate is calculated based on the following formula:

[0082] ;

[0083] Where D_desired is the desired average distance between vehicles (a pre-set distance, for example, 8-12 meters). Dis is the distance between two targets in a stop-and-go state, or the distance between a target in a stop-and-go state and the stop line.

[0084] Compare N_calculate with the system's preset maximum allowed interpolation target number N_max (e.g., N_max=2). This step is a critical safety redundancy design to prevent over-interpolation from causing a significant deviation between the virtual and real scenes.

[0085] The final number of interpolations, N_interpolate, is determined based on the following formula:

[0086] .

[0087] Optionally, after generating the virtual target corresponding to the lost target, the following steps are also included:

[0088] Configure the motion attribute of the virtual target to a walking / stopping state and add a virtual target identifier.

[0089] In this embodiment, all newly generated virtual targets are initially configured with a stop-and-go motion attribute and are assigned a virtual target identifier. This identifier is used to distinguish between real and virtual targets in subsequent processes and to perform differentiated track management.

[0090] Optionally, after generating the virtual target corresponding to the lost target, the following steps are also included:

[0091] If the traffic is congested and the target in a stop-and-go state meets the deletion criteria, then the virtual target in the stop-and-go state is deleted. The deletion criteria include: the relative distance between the virtual target's track and each moving track in the track list is less than a sixth distance threshold.

[0092] Optional, also includes:

[0093] If the traffic is not congested, delete the virtual targets that have continuously lost a preset number of point cloud frames and are in a stop-and-go state.

[0094] In this embodiment, the preset number can be a pre-defined number. The deleted virtual target needs to continuously lose a preset number of point cloud frames and be in a stop-and-go state.

[0095] In this embodiment, virtual target deletion involves traversing the track list and calculating the velocity V of each moving track. O >Speed ​​threshold V T The relative distance between the moving track and the target track (Distance) is considered if the moving track is close to the target track (Distance < distance threshold D). TIf the virtual target loses M frames consecutively in a non-congested scenario, the deletion logic will be triggered.

[0096] Optional, also includes:

[0097] If a point cloud cluster exists that is associated with a virtual target, then delete the maintenance identifier or the virtual target identifier.

[0098] It should be noted that virtual targets are crucial for maintaining stability, but improper management can lead to risks such as track delays and correlation conflicts. To mitigate these risks, track management of virtual targets is necessary, and the specific steps are as follows:

[0099] Virtual target deletion: Traverse the track list and calculate the velocity V of each moving track. O >Speed ​​threshold V T The relative distance between the moving track and the target track (Distance) is considered if the moving track is close to the target track (Distance < distance threshold D). T If the virtual target loses M frames consecutively in a non-congested scenario, the deletion logic will be triggered.

[0100] Virtual target conversion: If a point cloud cluster associated with a virtual target exists, the maintenance identifier or virtual target identifier is deleted. Thereafter, the track is tracked and maintained as a normal track.

[0101] Virtual target activation: For virtual targets, the association threshold is relaxed to prevent the real and virtual positions of the target from not matching when the target is activated, resulting in the inability to associate with real detection points and causing track delays or track redundancy.

[0102] Optional, also includes:

[0103] Point clouds that are in the same lane and whose longitudinal distance is less than the seventh distance threshold are clustered to obtain point cloud clusters;

[0104] The point cloud clusters are associated with the tracks in the track list, and the track list is updated based on the associated tracks.

[0105] In this embodiment, point clouds within the same lane and with similar longitudinal distances are preferentially clustered into a single target, significantly improving lateral resolution and fundamentally suppressing the mutual pulling of cross-lane targets. The point cloud clusters are then associated with tracks in the track list; for example, each confirmed track is assigned a "main lane" attribute. This attribute is determined based on the matching relationship between historical tracks and lane models. When calculating the association cost matrix, "lane consistency cost" is introduced as the highest priority weighting factor. If a newly detected target cluster shares the same "main lane" as a track, the association cost between them will be significantly reduced. If the lanes are different, the association cost will be significantly increased. The system preferentially selects pairs with consistent lanes and the lowest overall cost (distance, speed, lane) for association.

[0106] In this embodiment, the overall cost is determined based on the distance difference and velocity difference between the target trajectory and the point cloud cluster, where the distance difference is denoted as disRange and the velocity difference is denoted as disVelocity. The overall cost, Score, is calculated based on the following formula: Score = w1 × disRange + w2 × disVelocity, where w1 and w2 are preset values, w1 > w2, and w1 + w2 = 1, for example: w1 = 0.7, w2 = 0.3.

[0107] It should be noted that in congested scenarios, for stable (long lifecycle) longitudinally moving targets, the lateral velocity and lateral acceleration of the target are corrected by using the velocity information and displacement velocity information of the targets in front and behind, and the maximum allowable velocity fluctuation range of the target is limited, thereby further improving the trajectory stability in high-density target scenarios.

[0108] In this embodiment, under congested conditions, the velocity generated by filters such as Kalman filters / extended Kalman filters has poor accuracy, and the target's velocity and heading angle fluctuate significantly. Further correction of the target's velocity in this scenario is needed to improve the stability of the trajectory. The specific strategy is as follows: For stable targets (long lifespan age, such as age>15 frames, meaning a target that has existed for 15 frames), it is necessary to determine whether the target is a longitudinally moving target based on its own motion speed and the motion speeds of vehicles in front and behind. If it is a longitudinally moving target (the longitudinal velocity Vy of the target is much greater than the lateral velocity Vx, such as Vy>3Vx), then the lateral velocity Vx of the target is corrected. For targets in the initial stage (such as age<10), the lateral velocity of the target is corrected using the displacement velocity Vdx of the target. Vx=MIN(Vdx,Vx), Vdx=(xk-x0) / (k×T); xk is the position of the target in the kth frame, x0 is the position of the target in the initial frame, and T is the frame period. In addition, the maximum allowable fluctuation range of the target's lateral velocity and lateral acceleration is limited (such as limiting the maximum fluctuation range of lateral velocity and lateral acceleration between two consecutive frames).

[0109] In existing technologies, millimeter-wave radar suffers from insufficient resolution at long distances or for distinguishing lateral targets due to its inherent limited angular resolution. In congested traffic, the radar point clouds reflected by vehicles in adjacent lanes or closely spaced vehicles within the same lane highly overlap, almost forming a cluster. Traditional Euclidean distance-based clustering algorithms struggle to effectively separate point clouds belonging to different vehicles. Subsequently, tracking algorithms based on simple association strategies such as nearest neighbor or global nearest neighbor fail to correctly associate multiple targets due to the high similarity of their state vectors, leading to confusion and frequent ID jumps in the generated tracking trajectories among multiple real targets. This results in highly unstable output trajectory IDs, severely reducing tracking accuracy and reliability, and failing to provide continuous and reliable environmental perception information for downstream traffic condition assessment, event detection, or vehicle decision-making and control modules.

[0110] To address the aforementioned issues, this embodiment proposes clustering point clouds that are within the same lane and whose longitudinal distance is less than the seventh distance threshold to obtain point cloud clusters. These point cloud clusters are then associated with tracks in the track list, and the track list is updated based on the associated tracks. Through lane-constrained clustering and association, the "pulling" of trajectories and ID switching between multiple targets is greatly reduced, enabling the output of a stable and reliable target list, thereby solving the problem of unstable multi-target tracking.

[0111] S140, target tracking is performed based on the virtual target corresponding to the lost target and multiple targets.

[0112] In this embodiment, target tracking based on the virtual target corresponding to the lost target and multiple targets can be achieved by: sequentially inserting the virtual target corresponding to the lost target along the line connecting two targets in a stop-and-go state, based on target intervals; updating the track list based on multiple targets and virtual targets; and thus achieving target tracking. Alternatively, target tracking can be achieved by: sequentially inserting the virtual target corresponding to the lost target along the line connecting a target in a stop-and-go state and a stop line, based on target intervals; updating the track list based on multiple targets and virtual targets; and thus achieving target tracking.

[0113] Optionally, target tracking is performed based on the virtual target corresponding to the lost target and multiple targets, including:

[0114] On the line connecting two targets in a moving or stopped state, virtual targets corresponding to the lost target are sequentially inserted based on the target interval, and target tracking is performed based on multiple targets and virtual targets.

[0115] In a specific example, if two stable moving targets (Car_A and Car_B, assuming Car_A is in front of Car_B) are identified in the same lane, and the spatial distance Dis between the two targets is greater than a first threshold T1 (e.g., T1 = 15 meters) and less than a second threshold T2 (e.g., T2 = 50 meters), then N_interpolate virtual targets are inserted sequentially at intervals D_desired along the line connecting Car_A and Car_B. For example, when inserting a virtual target, its position is D_desired meters behind Car_A.

[0116] or,

[0117] On the line connecting the target in a moving-stop state and the stop line, virtual targets corresponding to the lost target are sequentially inserted based on the target interval, and target tracking is performed based on multiple targets and virtual targets.

[0118] In another specific example, a stable moving target (lead vehicle) is identified, with no other targets in front of it, but its current position is more than the threshold T (e.g., T=40 meters) away from the stop line. On the line connecting the lead vehicle and the stop line, virtual targets are inserted sequentially at intervals D_desired, starting from the stop line and moving backward.

[0119] In this embodiment, local microscopic traffic environment identification of the target vehicle is also required. Rear target presence identification: Using the center of the target vehicle as a reference, a detection area is formed by extending a certain distance backward within its lane. Other tracked targets are indexed to check if any target falls into this detection area. If so, the index ID of the rear target is recorded. Forward target presence identification: Using the center of the target vehicle as a reference, a detection area is formed by extending a certain distance forward within its lane, with the upper limit of the extension distance being the stop line. Other tracked targets are indexed to check if any target falls into this detection area. If so, the index ID of the forward target is recorded.

[0120] It's important to note that after inserting virtual targets, the indices of the preceding and following targets also need to be updated. For example: if two stable moving targets (CarA and CarB, assuming CarA is ahead of CarB) are identified in the same lane, and the spatial distance Dis between the two targets is greater than a first threshold T1 (e.g., T1 = 15 meters) and less than a second threshold T2 (e.g., T2 = 50 meters), then N_interpolate virtual targets (virtual target 1, virtual target 2, ...) are inserted sequentially at intervals D_desired along the line connecting CarA and CarB. Taking the insertion of virtual target 1 as an example: modify the following vehicle index of CarA: change it from pointing to CarB to pointing to V1. Set the preceding vehicle index of V1 to CarA and the following vehicle index to CarB. Modify the preceding vehicle index of CarB: change it from pointing to CarA to pointing to virtual target 1. At this point, the new logical vehicle chain becomes: CarA - Virtual Target 1 - CarB. Subsequent decisions can be made based on this updated chain relationship.

[0121] The technical solution of this embodiment receives point cloud data sent by roadside equipment; identifies the point cloud data to obtain multiple targets and traffic conditions; if the traffic condition is congested and there are lost targets, a virtual target corresponding to the lost target is generated; target tracking is performed based on the virtual target corresponding to the lost target and multiple targets, which can solve the problem of losing stationary or low-speed targets when the traffic condition is congested. When the traffic condition is congested, generating a virtual target of the lost target and performing target tracking based on the virtual target and multiple targets can improve the accuracy of target tracking.

[0122] Example 2

[0123] Figure 2 This is a schematic diagram of a target tracking device provided in an embodiment of the present invention. This embodiment is applicable to target tracking applications. The device can be implemented using software and / or hardware, and can be integrated into any device that provides target tracking functionality, such as… Figure 2 As shown, the target tracking device specifically includes: a point cloud data receiving module 210, a point cloud data recognition module 220, a virtual target generation module 230, and a target tracking module 240.

[0124] Among them, the point cloud data receiving module is used to receive point cloud data sent by roadside equipment;

[0125] The point cloud data recognition module is used to recognize the point cloud data to obtain multiple targets and traffic conditions;

[0126] The virtual target generation module is used to generate virtual targets corresponding to lost targets if the traffic status is congested and lost targets exist.

[0127] The target tracking module is used to track targets based on the virtual target corresponding to the lost target and multiple targets.

[0128] The above-described products can perform the methods provided in any embodiment of the present invention, and have the corresponding functional modules and beneficial effects for performing the methods.

[0129] Example 3

[0130] Figure 3 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0131] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0132] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0133] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as target tracking methods.

[0134] In some embodiments, the target tracking method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the target tracking method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the target tracking method by any other suitable means (e.g., by means of firmware).

[0135] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0136] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0137] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0138] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0139] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0140] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0141] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0142] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the target tracking method according to any embodiment of the invention.

[0143] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0144] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A target tracking method, characterized in that, include: Receive point cloud data sent by roadside equipment; The point cloud data is then identified to obtain multiple targets and traffic conditions; If the traffic is congested and there is a lost target, then a virtual target corresponding to the lost target is generated. Target tracking is performed based on the virtual target corresponding to the lost target and multiple targets.

2. The method according to claim 1, characterized in that, If the traffic situation is congested and there are lost targets, then generate virtual targets corresponding to the lost targets, including: If the traffic is congested, there is a lost target, and there are no other targets in front of the lost target but there are targets behind it, then a virtual target corresponding to the lost target is generated based on the movement speed of the targets behind the lost target. This continues until the distance between the generated virtual target corresponding to the lost target and the stop line is less than the first distance threshold. Then, the movement attribute of the virtual target is changed from a moving state to a moving-stopping state, and a maintenance flag is added. If the traffic is congested, there is a lost target, and there are no other targets in front of or behind the lost target, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target. This continues until the distance between the generated virtual target corresponding to the lost target and the stop line is less than the first distance threshold. Then, the movement attribute of the virtual target is changed from a moving state to a moving-stopping state, and a maintenance flag is added. If the traffic is congested, there is a lost target, and the target in front of the lost target is in the first state, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target. This continues until the distance between the generated virtual target corresponding to the lost target and the target in front of the lost target is less than the second distance threshold. Then, the movement attribute of the virtual target is changed from the moving state to the walking-stopping state, and a maintenance flag is added. If the traffic is congested, there is a lost target, and the target in front of the lost target is in the second state, then a virtual target corresponding to the lost target is generated based on the historical movement speed of the lost target and the movement speed of the target in front of the lost target.

3. The method according to claim 2, characterized in that, Based on the historical speed of the lost target and the speed of targets ahead of it, a virtual target corresponding to the lost target is generated, including: The target's speed is obtained by weighted summing the historical speed of the lost target and the speed of the target ahead of the lost target. A virtual target corresponding to the lost target is generated based on the target's movement speed.

4. The method according to claim 1, characterized in that, If the traffic situation is congested and there are lost targets, before generating the virtual target corresponding to the lost target, the following steps are also included: If there are two targets in a stop-and-go state within the same lane, and the spatial distance between the two targets is greater than a third distance threshold and less than a fourth distance threshold, wherein the third distance threshold is less than the fourth distance threshold, Alternatively, if there are no other targets in front of the target in a stopped state, and the distance between the target in a stopped state and the stop line is greater than the fifth distance threshold, This confirms that a target has been lost.

5. The method according to claim 4, characterized in that, If the traffic situation is congested and there are lost targets, then generate virtual targets corresponding to the lost targets, including: If the traffic is congested and there are lost targets, the interpolation quantity is determined based on the distance between two targets in a stop-and-go state, and a virtual target corresponding to the lost target is generated based on the interpolation quantity. or, If the traffic is congested and there are lost targets, the number of interpolations is determined based on the distance between the target in the stop-and-go state and the stop line, and a virtual target corresponding to the lost target is generated based on the number of interpolations.

6. The method according to claim 5, characterized in that, Target tracking is performed based on the virtual target corresponding to the lost target and multiple targets, including: On the line connecting two targets that are in a moving or stopped state, virtual targets corresponding to the lost targets are inserted sequentially based on the target interval, and target tracking is performed based on multiple targets and virtual targets; or, On the line connecting the target in a moving-stop state and the stop line, virtual targets corresponding to the lost target are sequentially inserted based on the target interval, and target tracking is performed based on multiple targets and virtual targets.

7. The method according to claim 5, characterized in that, After generating the virtual target corresponding to the lost target, the process also includes: Configure the motion attribute of the virtual target to a walking / stopping state and add a virtual target identifier.

8. The method according to claim 7, characterized in that, After generating the virtual target corresponding to the lost target, the process also includes: If the traffic is congested and the target in a stop-and-go state meets the deletion criteria, then the virtual target in the stop-and-go state is deleted. The deletion criteria include: the relative distance between the virtual target's track and each moving track in the track list is less than a sixth distance threshold.

9. The method according to claim 8, characterized in that, Also includes: If the traffic is not congested, delete the virtual targets that have continuously lost a preset number of point cloud frames and are in a stop-and-go state.

10. The method according to claim 1, characterized in that, Also includes: If a point cloud cluster exists that is associated with a virtual target, then delete the maintenance identifier or the virtual target identifier.

11. The method according to claim 1, characterized in that, Also includes: Point clouds that are in the same lane and whose longitudinal distance is less than the seventh distance threshold are clustered to obtain point cloud clusters; The point cloud clusters are associated with the tracks in the track list, and the track list is updated based on the associated tracks.

12. A target tracking device, characterized in that, include: The point cloud data receiving module is used to receive point cloud data sent by roadside equipment; The point cloud data recognition module is used to recognize the point cloud data to obtain multiple targets and traffic conditions; The virtual target generation module is used to generate virtual targets corresponding to lost targets if the traffic status is congested and lost targets exist. The target tracking module is used to track targets based on the virtual target corresponding to the lost target and multiple targets.

13. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the target tracking method according to any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the target tracking method according to any one of claims 1-11.

15. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the target tracking method according to any one of claims 1-11.

Images