A stationary vehicle target recognition method, device, equipment and storage medium

Abstract

A stationary vehicle target recognition method, device, equipment and storage medium, relating to the technical field of signal processing and target tracking, the stationary vehicle target recognition method comprises: separating the acquired point cloud data into static and dynamic data, performing guardrail fitting and recognition based on all static point clouds, and determining the guardrail fitting curve; judging whether the stationary target scene beside the guardrail is satisfied according to the vehicle motion information and the guardrail fitting curve; when the stationary target scene beside the guardrail is satisfied, screening out the target static point cloud that meets the pitch angle requirement and is outside the guardrail fitting curve and close to the vehicle side, adjusting the confidence of the guardrail static track and controlling the start of the guardrail static track through point cloud association, point cloud clustering and point cloud data transmission; updating and deleting the guardrail static track according to the confidence to identify the stationary vehicle target. The present invention can effectively solve the problem that the millimeter wave radar cannot identify the stationary vehicle target in front when there are interference objects beside it.

Description

Technical Field

[0001] The present application relates to the technical field of signal processing and target tracking, and in particular to a method, device, equipment and storage medium for identifying a stationary vehicle target. Background Art

[0002] With the continuous improvement of automobile electronics and intelligence, advanced driving assistance functions are increasingly valued by major automobile manufacturers and consumers. This function plays an important role in improving vehicle driving safety.

[0003] As an all-weather road target detection sensor, millimeter wave radar has the characteristics of high operating frequency, short wavelength, small antenna size, and strong environmental adaptability. It can effectively detect target distance, relative speed, direction and other information, and can provide stable sensor target-level data for the development of advanced driver assistance system functions.

[0004] However, when detecting stationary targets, millimeter-wave radar has unstable tracking due to uneven point distribution, a small number of points, and the continuity of the detection points is greatly affected by the vehicle speed. It is also basically unable to identify the size and type of stationary objects. Especially for conditions such as red lights and parked vehicles next to the guardrail in front of the intersection, there is a common problem that millimeter-wave radar cannot recognize stationary vehicle targets in front when there are interference objects next to them. Summary of the invention

[0005] The present application provides a method, device, equipment and storage medium for identifying a stationary vehicle target, which can effectively solve the problem that a millimeter wave radar cannot identify a stationary vehicle target in front when there are interference objects nearby.

[0006] In a first aspect, an embodiment of the present application provides a stationary vehicle target recognition method, the stationary vehicle target recognition method comprising:

[0007] Separate the acquired point cloud data into static and dynamic ones, perform guardrail fitting and recognition based on all static point clouds, and determine the guardrail fitting curve;

[0008] Based on the vehicle motion information and the guardrail fitting curve, determine whether the static target scenario next to the guardrail is met;

[0009] When the scene of a stationary target next to a guardrail is met, the target stationary point cloud that meets the pitch angle requirements and is outside the guardrail fitting curve and close to the vehicle side is selected. Through point cloud association, point cloud clustering and point cloud data transmission, the confidence of the guardrail stationary track is adjusted and the start of the guardrail stationary track is controlled.

[0010] The guardrail stationary track is updated and deleted according to the confidence level to identify the stationary vehicle target.

[0011] In combination with the first aspect, in one implementation, the step of separating the acquired point cloud data into static and dynamic data, performing guardrail fitting and identification based on all static point clouds, and determining the guardrail fitting curve includes:

[0012] Obtain millimeter-wave radar point cloud data, and separate the current frame point cloud into static and dynamic ones according to the speed threshold;

[0013] All stationary point clouds are fitted and identified using the least squares method to determine the guardrail fitting curve.

[0014] In combination with the first aspect, in one implementation, judging whether a static target scene beside a guardrail is satisfied according to the vehicle motion information and the guardrail fitting curve includes:

[0015] Determine whether the vehicle is in a turning state based on the vehicle speed and yaw angle;

[0016] According to the guardrail fitting curve, determine whether the guardrail is approximately a straight line and parallel to the vehicle's travel direction, and determine whether the guardrail is in the lane adjacent to the vehicle;

[0017] If the vehicle is not turning, the guardrail is approximately a straight line and parallel to the direction of travel of the vehicle, and the guardrail is in the lane adjacent to the vehicle, then the stationary target scenario next to the guardrail is determined to be met.

[0018] In combination with the first aspect, in one implementation, judging whether the guardrail is approximately a straight line and parallel to the direction of travel of the vehicle according to the guardrail fitting curve, and judging whether the guardrail is in the lane adjacent to the vehicle, includes:

[0019] When the coefficient of the guardrail fitting curve is between -0.2 and 0.2, it is determined that the guardrail is approximately a straight line and parallel to the vehicle's travel direction;

[0020] When the constant term of the guardrail fitting curve is between 0 and 3.6, it is determined that the guardrail is in the adjacent lane of the vehicle.

[0021] In combination with the first aspect, in one implementation, the step of adjusting the confidence of the guardrail stationary track and controlling the start of the guardrail stationary track through point cloud association, point cloud clustering, and point cloud data transmission includes:

[0022] According to the distance between the target stationary point cloud and the vehicle, different clustering starting thresholds are set, and it is determined whether there is a guardrail stationary track at present;

[0023] When a guardrail static track already exists:

[0024] If there is a confirmed track by CFD, point cloud association is performed according to the track position and the clustering radius confirmed in the previous frame. If the association is successful, the confidence level remains unchanged. If no point cloud association is successful, the confidence level is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold.

[0025] If there is a TBC track to be confirmed, the point cloud is associated according to the clustering radius confirmed in the previous frame. If the association is successful, the confidence is increased. If no point cloud is successfully associated, the confidence is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold;

[0026] When there is no guardrail static track, the target static point cloud is directly clustered according to the clustering starting threshold;

[0027] If clustering is successful, the CFD track is started, the confidence is set to the first set value, and the center point and cluster radius of successful clustering are recorded;

[0028] If clustering is unsuccessful, determine whether there is a previous frame to pass the point cloud data to be processed. If so, cluster the previous frame and the current frame to be processed point cloud data. If clustering is successful, record the cluster center point and cluster radius, mark the track status as TBC, and set the confidence to the second set value; if not, predict the position of the point cloud in the radar coordinate system at the next moment according to the vehicle's motion state, and pass it to the next frame of point cloud data to be processed.

[0029] In combination with the first aspect, in one implementation, updating and deleting the guardrail stationary track according to the confidence level includes:

[0030] After the initial CFD track, if the confidence is lower than the second set value, the track is deleted;

[0031] When the TBC track confidence is lower than the third set value, the track is deleted. When the TBC track confidence is higher than the first set value, the track state is changed from TBC to CFD, and the CFD track is started.

[0032] In combination with the first aspect, in one implementation, the method further includes:

[0033] The point cloud data to be processed is numbered, and the number is increased by 1 each time it is passed until the maximum number is reached and the transfer stops.

[0034] When the point cloud data to be processed in the previous frame and the current frame are clustered successfully, the point cloud data labels transmitted in the previous frame are modified to the maximum label.

[0035] In a second aspect, an embodiment of the present application provides a stationary vehicle target recognition device, the stationary vehicle target recognition device comprising:

[0036] A data processing module is used to separate the acquired point cloud data into static and dynamic ones, perform guardrail fitting and identification based on all static point clouds, and determine the guardrail fitting curve;

[0037] The scene recognition module determines whether the scene of a stationary target next to the guardrail is met based on the vehicle motion information and the guardrail fitting curve;

[0038] The track initiation module is used to select the target static point cloud that meets the pitch angle requirement and is close to the vehicle side outside the guardrail fitting curve when the static target scene next to the guardrail is met, and adjust the confidence of the guardrail static track and control the start of the guardrail static track through point cloud association, point cloud clustering and point cloud data transmission;

[0039] A track tracking module updates and deletes the static track of the guardrail according to the confidence level to identify the static vehicle target.

[0040] In a third aspect, an embodiment of the present application provides a stationary vehicle target recognition device, which includes a processor, a memory, and a stationary vehicle target recognition program stored in the memory and executable by the processor, wherein when the stationary vehicle target recognition program is executed by the processor, the steps of the stationary vehicle target recognition method as described in any one of the above are implemented.

[0041] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a stationary vehicle target recognition program is stored, wherein when the stationary vehicle target recognition program is executed by a processor, the steps of any of the stationary vehicle target recognition methods described above are implemented.

[0042] The beneficial effects brought by the technical solution provided by the embodiment of the present application include at least:

[0043] The stationary vehicle target recognition method in the present application separates the acquired point cloud data into static and dynamic data, performs guardrail fitting and recognition based on all static point clouds, and determines the guardrail fitting curve; judges whether the stationary target scenario next to the guardrail is met based on the vehicle motion information and the guardrail fitting curve; when the stationary target scenario next to the guardrail is met, screens out the target stationary point cloud that meets the pitch angle requirements and is outside the guardrail fitting curve and close to the vehicle side, and adjusts the confidence of the guardrail stationary track and controls the start of the guardrail stationary track through point cloud association, point cloud clustering and point cloud data transmission; updates and deletes the guardrail stationary track according to the confidence to identify the stationary vehicle target.

[0044] This application comprehensively considers the clustering starting conditions of the point cloud level under specific environments, the point track screening and processing in specific areas, and the multi-frame accumulation at the point cloud level, and adds confidence-related judgment logic. The stationary track of the guardrail is updated and deleted according to the size of the confidence level, thereby effectively solving the problem of the millimeter-wave radar not being able to recognize the stationary vehicle target in front when there are interference objects next to it. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of an embodiment of a method for identifying a stationary vehicle target of the present application;

[0046] Figure 2 This is a schematic diagram of the functional modules of an embodiment of a stationary vehicle target recognition device of the present application;

[0047] Figure 3 This is a schematic diagram of the hardware structure of a stationary vehicle target recognition device involved in the embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0049] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit "first", "second" and "third" to different types.

[0050] In the description of the embodiments of the present application, "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a specific way.

[0051] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; the “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0052] In some processes described in the embodiments of the present application, multiple operations or steps that appear in a specific order are included, but it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or in parallel, and the sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0053] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0054] In a first aspect, an embodiment of the present application provides a method for identifying a stationary vehicle target.

[0055] In one embodiment, referring to Figure 1 , Figure 1 This is a flow chart of an embodiment of a method for identifying a stationary vehicle target of the present application. Figure 1 As shown, the stationary vehicle target recognition method includes:

[0056] S1. Separate the acquired point cloud data into static and dynamic ones, perform guardrail fitting and recognition based on all static point clouds, and determine the guardrail fitting curve;

[0057] Specifically, step S1 includes:

[0058] S11, obtaining millimeter wave radar point cloud data, and separating the current frame point cloud into static and dynamic points according to the speed threshold;

[0059] In this embodiment, for the point cloud data of the millimeter-wave radar, whether it is a stationary point cloud is determined based on the set speed threshold. That is, if the speed of the current frame point cloud is less than the set threshold, it is classified as a stationary point cloud, thereby performing dynamic and static separation.

[0060] S12. All stationary point clouds are fitted and identified using the least square method to determine a guardrail fitting curve.

[0061] The least square method is used to perform curve fitting in this embodiment. Of course, other methods can be used for fitting as needed, and this embodiment does not limit this.

[0062] S2, judging whether the static target scenario beside the guardrail is met based on the vehicle motion information and the guardrail fitting curve;

[0063] Specifically, step S2 includes:

[0064] S21, judging whether the vehicle is in a turning state according to the vehicle speed and the yaw angle;

[0065] In specific implementation, for example, when the vehicle speed is ≥a Km / h and the yaw angle is ≤brad / s, it is determined that the vehicle is not in a turning state.

[0066] S22, judging whether the guardrail is approximately a straight line and parallel to the direction of travel of the vehicle according to the guardrail fitting curve, and judging whether the guardrail is in the lane adjacent to the vehicle;

[0067] The guardrail fitting curve can be expressed as: y = f (x) = c3 * x 3 +c2*x 2 +c1*x+c0, determine whether the guardrail curvatures c3, c2, and c1 are between -0.2 and 0.2. If so, the guardrail is determined to be approximately a straight line and parallel to the vehicle's travel direction). In addition, for the distance between the guardrail and the edge of the vehicle body, that is, determine whether the c0 parameter is between 0m and 3.6m. If so, the guardrail is determined to be in the adjacent lane of the vehicle.

[0068] S23: If the vehicle is not in a turning state, the guardrail is approximately a straight line and parallel to the direction of travel of the vehicle, and the guardrail is in the lane adjacent to the vehicle, it is determined that the stationary target scenario next to the guardrail is met.

[0069] It can be understood that when the conditions in steps S21 and S22 are both met, it is determined that the stationary target scene next to the guardrail is met.

[0070] S3. When the scene of a stationary target next to a guardrail is met, the target stationary point cloud that meets the pitch angle requirement and is outside the guardrail fitting curve and close to the vehicle side is selected, and the confidence of the guardrail stationary track is adjusted and the start of the guardrail stationary track is controlled through point cloud association, point cloud clustering and point cloud data transmission;

[0071] It is worth noting that the track start is the generation of a new track. The generation of a new track is mainly through processing the remaining point tracks obtained by multiple radar detections to extract potential target tracks. For target tracks that cannot be updated for a long time, they should be removed from the track library, that is, the track is terminated. The track termination is conducive to maintaining the capacity of the radar detection track library and deleting invalid target information in time, so that the latest target information is retained. For the track start, its basic function is to extract the target initial track information from the point track information detected by the radar.

[0072] In step S3, the static point clouds are first screened, that is, those point clouds that are not involved in the guardrail fitting, are located close to the vehicle, and have a pitch angle that meets the requirements (confirmed to be low-altitude obstacle point clouds) are selected. Then, the classification is explained based on whether there is a guardrail static track. Specifically:

[0073] S31, setting different clustering starting thresholds according to the distance between the target stationary point cloud and the vehicle, and determining whether a guardrail stationary track currently exists;

[0074] For example, when the stationary point cloud is in the Y1 (0-40m) area, the clustering starting threshold is set to the first threshold (N1, R1); when the stationary point cloud is in the Y2 (40-80m) area, the clustering starting threshold is set to the first threshold (N2, R2); when the stationary point cloud is in the Y2 (above 80m) area, the clustering starting threshold is set to the third threshold (N3, R3). Where N is the center point and R is the clustering radius. It can be understood that multiple thresholds corresponding to different ranges can be set according to actual needs, and this embodiment does not limit this.

[0075] S32: When a guardrail static track already exists:

[0076] If there is a confirmed track by CFD, point cloud association is performed according to the track position and the clustering radius confirmed in the previous frame. If the association is successful, the confidence level remains unchanged. If no point cloud association is successful, the confidence level is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold.

[0077] Successful association means that a certain number of point clouds have been reached, and both the previous frame and the current frame are static and within the cluster radius. In specific implementation, if the association is successful, the confidence remains unchanged, and if no point cloud is successfully associated, the confidence is -1.

[0078] If there is a TBC track to be confirmed, the point cloud is associated according to the clustering radius confirmed in the previous frame. If the association is successful, the confidence is increased. If no point cloud is successfully associated, the confidence is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold;

[0079] In specific implementation, if the point cloud is successfully associated, the TBC track confidence is increased by 1; if no point cloud is successfully associated, the confidence is decreased by 1.

[0080] S33, when there is no guardrail static track, the target static point cloud is directly clustered according to the clustering starting threshold;

[0081] If clustering is successful, the CFD track is started, the confidence level is set to the first set value, and the center point and cluster radius of successful clustering are recorded; for example, if clustering is successful, the confidence level can be set to 10.

[0082] If clustering is unsuccessful, determine whether there is a previous frame to pass the point cloud data to be processed. If so, cluster the previous frame and the current frame to be processed point cloud data. If clustering is successful, record the cluster center point and cluster radius, mark the track status as TBC, and set the confidence to the second set value; if not, predict the position of the point cloud in the radar coordinate system at the next moment according to the vehicle's motion state, and pass it to the next frame of point cloud data to be processed.

[0083] Specifically, the point cloud data to be processed are labeled, and the number is increased by 1 each time it is passed until the maximum number is reached and the transmission is stopped.

[0084] When the point cloud data to be processed in the previous frame and the current frame are clustered successfully, the point cloud data labels transmitted in the previous frame are modified to the maximum label.

[0085] In this embodiment, labels 1, 2, and 3 are set, and label 3 is the largest label. The point cloud data to be processed in the previous frame can be labeled 1, 2, and 3. When clustering is successful, the labels of the point cloud data transmitted in the previous frame with successful clustering will be modified to 3. The point cloud data is transmitted up to 3 times, and the data with label 4 is no longer transmitted. It can be understood that the number of transmissions can also be set to other sizes as needed, and this embodiment does not limit it here.

[0086] S4. Update and delete the guardrail stationary track according to the confidence level to identify the stationary vehicle target.

[0087] In specific implementation, after the CFD track is started, if the confidence is lower than the second set value, the track is deleted; when the TBC track confidence is lower than the third set value, the track is deleted; when the TBC track confidence is higher than the first set value, the track state is changed from TBC to CFD, and the CFD track is started.

[0088] For example, for a successfully initiated stationary target CFD track, if the confidence is lower than 5, the track is deleted. For a TBC track, if the confidence is lower than 0, the track is deleted. If the confidence is higher than 10, the track status changes from TBC to CFD, and the CFD track is initiated.

[0089] In summary, the stationary vehicle target recognition method in the present application separates the acquired point cloud data into static and dynamic data, performs guardrail fitting and recognition based on all static point clouds, and determines the guardrail fitting curve; judges whether the stationary target scenario next to the guardrail is met based on the vehicle motion information and the guardrail fitting curve; when the stationary target scenario next to the guardrail is met, screens out the target stationary point cloud that meets the pitch angle requirements and is outside the guardrail fitting curve and close to the vehicle side, and adjusts the confidence of the guardrail stationary track and controls the starting of the guardrail stationary track through point cloud association, point cloud clustering and point cloud data transmission; updates and deletes the guardrail stationary track according to the confidence to identify the stationary vehicle target.

[0090] This application comprehensively considers the clustering starting conditions of the point cloud level under specific environments, the point track screening and processing in specific areas, and the multi-frame accumulation at the point cloud level, and adds confidence-related judgment logic. The stationary track of the guardrail is updated and deleted according to the size of the confidence level, thereby effectively solving the problem of the millimeter-wave radar not being able to recognize the stationary vehicle target in front when there are interference objects next to it.

[0091] In a second aspect, an embodiment of the present application also provides a stationary vehicle target recognition device.

[0092] In one embodiment, referring to Figure 2 , Figure 2 This is a functional module diagram of an embodiment of a stationary vehicle target recognition device of the present application. Figure 2 As shown, the stationary vehicle target recognition device includes:

[0093] A data processing module is used to separate the acquired point cloud data into static and dynamic ones, perform guardrail fitting and identification based on all static point clouds, and determine the guardrail fitting curve;

[0094] The scene recognition module determines whether the scene of a stationary target next to the guardrail is met based on the vehicle motion information and the guardrail fitting curve;

[0095] The track initiation module is used to select the target static point cloud that meets the pitch angle requirement and is close to the vehicle side outside the guardrail fitting curve when the static target scene next to the guardrail is met, and adjust the confidence of the guardrail static track and control the start of the guardrail static track through point cloud association, point cloud clustering and point cloud data transmission;

[0096] A track tracking module updates and deletes the static track of the guardrail according to the confidence level to identify the static vehicle target.

[0097] Furthermore, in one embodiment, the data processing module separates the acquired point cloud data into static and dynamic data, performs guardrail fitting and recognition based on all static point clouds, and determines the guardrail fitting curve, including:

[0098] Obtain millimeter-wave radar point cloud data, and separate the current frame point cloud into static and dynamic ones according to the speed threshold;

[0099] All stationary point clouds are fitted and identified using the least squares method to determine the guardrail fitting curve.

[0100] Furthermore, in one embodiment, the scene recognition module determines whether the scene of a stationary target next to a guardrail is satisfied based on the vehicle motion information and the guardrail fitting curve, including:

[0101] Determine whether the vehicle is in a turning state based on the vehicle speed and yaw angle;

[0102] According to the guardrail fitting curve, determine whether the guardrail is approximately a straight line and parallel to the vehicle's travel direction, and determine whether the guardrail is in the lane adjacent to the vehicle;

[0103] If the vehicle is not turning, the guardrail is approximately a straight line and parallel to the direction of travel of the vehicle, and the guardrail is in the lane adjacent to the vehicle, then the stationary target scenario next to the guardrail is determined to be met.

[0104] Furthermore, in one embodiment, the scene recognition module determines whether the guardrail is approximately a straight line and parallel to the vehicle's traveling direction according to the guardrail fitting curve, and determines whether the guardrail is in the lane adjacent to the vehicle, including:

[0105] When the coefficient of the guardrail fitting curve is between -0.2 and 0.2, it is determined that the guardrail is approximately a straight line and parallel to the vehicle's travel direction;

[0106] When the constant term of the guardrail fitting curve is between 0 and 3.6, it is determined that the guardrail is in the adjacent lane of the vehicle.

[0107] Furthermore, in one embodiment, the track initiation module adjusts the confidence of the guardrail stationary track and controls the initiation of the guardrail stationary track through point cloud association, point cloud clustering and point cloud data transmission, including:

[0108] According to the distance between the target stationary point cloud and the vehicle, different clustering starting thresholds are set, and it is determined whether there is a guardrail stationary track at present;

[0109] When a guardrail static track already exists:

[0110] If there is a confirmed track by CFD, point cloud association is performed according to the track position and the clustering radius confirmed in the previous frame. If the association is successful, the confidence level remains unchanged. If no point cloud association is successful, the confidence level is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold.

[0111] If there is a TBC track to be confirmed, the point cloud is associated according to the clustering radius confirmed in the previous frame. If the association is successful, the confidence is increased. If no point cloud is successfully associated, the confidence is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold;

[0112] When there is no guardrail static track, the target static point cloud is directly clustered according to the clustering starting threshold;

[0113] If clustering is successful, the CFD track is started, the confidence is set to the first set value, and the center point and cluster radius of successful clustering are recorded;

[0114] If clustering is unsuccessful, determine whether there is a previous frame to pass the point cloud data to be processed. If so, cluster the previous frame and the current frame to be processed point cloud data. If clustering is successful, record the cluster center point and cluster radius, mark the track status as TBC, and set the confidence to the second set value; if not, predict the position of the point cloud in the radar coordinate system at the next moment according to the vehicle's motion state, and pass it to the next frame of point cloud data to be processed.

[0115] Furthermore, in one embodiment, the track tracking module updates and deletes the static track of the guardrail according to the confidence level, including:

[0116] After the initial CFD track, if the confidence is lower than the second set value, the track is deleted;

[0117] When the TBC track confidence is lower than the third set value, the track is deleted. When the TBC track confidence is higher than the first set value, the track state is changed from TBC to CFD, and the CFD track is started.

[0118] Furthermore, in one embodiment, the track initiation module is also used for:

[0119] The point cloud data to be processed is numbered, and the number is increased by 1 each time it is passed until the maximum number is reached and the transfer stops.

[0120] When the point cloud data to be processed in the previous frame and the current frame are clustered successfully, the point cloud data labels transmitted in the previous frame are modified to the maximum label.

[0121] Among them, the functional implementation of each module in the above-mentioned stationary vehicle target recognition device corresponds to the various steps in the above-mentioned stationary vehicle target recognition method embodiment, and its functions and implementation processes are no longer repeated here.

[0122] In a third aspect, an embodiment of the present application provides a stationary vehicle target recognition device, which may be a device with data processing capabilities, such as a personal computer (PC), a laptop computer, or a server.

[0123] Reference Figure 3 , Figure 3 Schematic diagram of the hardware structure of the stationary vehicle target recognition device involved in the embodiment of the present application. In the embodiment of the present application, the stationary vehicle target recognition device may include a processor, a memory, a communication interface and a communication bus.

[0124] The communication bus may be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0125] The communication interface includes an input / output (I / O) interface, a physical interface, and a logical interface, etc., which are used to interconnect the devices inside the stationary vehicle target recognition device, and an interface used to interconnect the stationary vehicle target recognition device with other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, an optical fiber interface, an ATM interface, etc.; the user device can be a display, a keyboard, etc.

[0126] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0127] The processor may be a general-purpose processor, and the general-purpose processor may call the stationary vehicle target recognition program stored in the memory and execute the stationary vehicle target recognition method provided by the embodiment of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the stationary vehicle target recognition program is called may refer to the various embodiments of the stationary vehicle target recognition method of the present application, and will not be repeated here.

[0128] Those skilled in the art will understand that Figure 3 The hardware structure shown in the figure does not constitute a limitation on the present application, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0129] In a fourth aspect, an embodiment of the present application also provides a computer-readable storage medium.

[0130] The computer-readable storage medium of the present application stores a stationary vehicle target recognition program, wherein when the stationary vehicle target recognition program is executed by a processor, the steps of the stationary vehicle target recognition method as described above are implemented.

[0131] Among them, the method implemented when the stationary vehicle target recognition program is executed can refer to the various embodiments of the stationary vehicle target recognition method of the present application, and will not be repeated here.

[0132] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0133] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, and includes a number of instructions for a terminal device to execute the methods described in each embodiment of the present application.

[0134] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A stationary vehicle target recognition method, characterized in that: The stationary vehicle target recognition method comprises: Separate the acquired point cloud data into static and dynamic ones, perform guardrail fitting and recognition based on all static point clouds, and determine the guardrail fitting curve; Based on the vehicle motion information and the guardrail fitting curve, determine whether the static target scenario next to the guardrail is met; When the scene of a stationary target next to a guardrail is met, the target stationary point cloud that meets the pitch angle requirements and is outside the guardrail fitting curve and close to the vehicle side is selected. Through point cloud association, point cloud clustering and point cloud data transmission, the confidence of the guardrail stationary track is adjusted and the start of the guardrail stationary track is controlled. updating and deleting the guardrail stationary track according to the confidence level to identify the stationary vehicle target; The step of adjusting the confidence of the guardrail stationary track and controlling the start of the guardrail stationary track through point cloud association, point cloud clustering and point cloud data transmission includes: According to the distance between the target static point cloud and the vehicle, different clustering starting thresholds are set, and it is determined whether there is a guardrail static track at present; When a guardrail static track already exists: If there is a confirmed track by CFD, point cloud association is performed according to the track position and the clustering radius confirmed in the previous frame. If the association is successful, the confidence level remains unchanged. If no point cloud association is successful, the confidence level is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold. If there is a TBC track to be confirmed, the point cloud is associated according to the clustering radius confirmed in the previous frame. If the association is successful, the confidence is increased. If no point cloud is successfully associated, the confidence is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold; When there is no guardrail static track, the target static point cloud is directly clustered according to the clustering starting threshold; If clustering is successful, the CFD track is started, the confidence is set to the first set value, and the center point and cluster radius of successful clustering are recorded; If clustering is unsuccessful, determine whether there is a previous frame to pass the point cloud data to be processed. If so, cluster the previous frame and the current frame to be processed point cloud data. If clustering is successful, record the cluster center point and cluster radius, mark the track status as TBC, and set the confidence to the second set value; if not, predict the position of the point cloud in the radar coordinate system at the next moment according to the vehicle's motion state, and pass it to the next frame of point cloud data to be processed.

2. The stationary vehicle target recognition method according to claim 1, characterized in that: The step of separating the acquired point cloud data into static and dynamic ones, performing guardrail fitting and identification based on all static point clouds, and determining the guardrail fitting curve includes: Obtain millimeter-wave radar point cloud data, and separate the current frame point cloud into static and dynamic ones according to the speed threshold; All stationary point clouds are fitted and identified using the least squares method to determine the guardrail fitting curve.

3. The stationary vehicle target recognition method according to claim 1, characterized in that: The determining whether a static target scenario next to a guardrail is satisfied based on the vehicle motion information and the guardrail fitting curve includes: Determine whether the vehicle is in a turning state based on the vehicle speed and yaw angle; According to the guardrail fitting curve, determine whether the guardrail is approximately a straight line and parallel to the vehicle's travel direction, and determine whether the guardrail is in the lane adjacent to the vehicle; If the vehicle is not turning, the guardrail is approximately a straight line and parallel to the direction of travel of the vehicle, and the guardrail is in the lane adjacent to the vehicle, then the stationary target scenario next to the guardrail is determined to be met.

4. The stationary vehicle target recognition method according to claim 3, characterized in that: The step of judging whether the guardrail is approximately a straight line and parallel to the vehicle's traveling direction according to the guardrail fitting curve, and judging whether the guardrail is in the lane adjacent to the vehicle, includes: When the coefficient of the guardrail fitting curve is between -0.2 and 0.2, it is determined that the guardrail is approximately a straight line and parallel to the vehicle's travel direction; When the constant term of the guardrail fitting curve is between 0 and 3.6, it is determined that the guardrail is in the adjacent lane of the vehicle.

5. The stationary vehicle target recognition method according to claim 1, characterized in that: The updating and deleting of the guardrail stationary track according to the confidence level includes: After the initial CFD track, if the confidence is lower than the second set value, the track is deleted; When the TBC track confidence is lower than the third set value, the track is deleted. When the TBC track confidence is higher than the first set value, the track state is changed from TBC to CFD, and the CFD track is started.

6. The stationary vehicle target recognition method according to claim 1, characterized in that: Also includes: The point cloud data to be processed is numbered, and the number is increased by 1 each time it is passed until the maximum number is reached and the transfer stops. When the point cloud data to be processed in the previous frame and the current frame are clustered successfully, the point cloud data labels transmitted in the previous frame are modified to the maximum label.

7. A stationary vehicle target recognition device, characterized in that: The stationary vehicle target recognition device comprises: A data processing module is used to separate the acquired point cloud data into dynamic and static ones, perform guardrail fitting and identification based on all static point clouds, and determine the guardrail fitting curve; The scene recognition module determines whether the scene of a static target next to the guardrail is met based on the vehicle motion information and the guardrail fitting curve; The track initiation module is used to select the target static point cloud that meets the pitch angle requirement and is close to the vehicle side outside the guardrail fitting curve when the static target scene next to the guardrail is met, and adjust the confidence of the guardrail static track and control the start of the guardrail static track through point cloud association, point cloud clustering and point cloud data transmission; A track tracking module, which updates and deletes the static track of the guardrail according to the confidence level to identify the static vehicle target; The device is also used for: According to the distance between the target static point cloud and the vehicle, different clustering starting thresholds are set, and it is determined whether there is a guardrail static track at present; When a guardrail static track already exists: If there is a confirmed track by CFD, point cloud association is performed according to the track position and the clustering radius confirmed in the previous frame. If the association is successful, the confidence level remains unchanged. If no point cloud association is successful, the confidence level is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold. If there is a TBC track to be confirmed, the point cloud is associated according to the clustering radius confirmed in the previous frame. If the association is successful, the confidence is increased. If no point cloud is successfully associated, the confidence is reduced, and the remaining target static point clouds that have not been successfully associated are clustered according to the clustering starting threshold; When there is no guardrail static track, the target static point cloud is directly clustered according to the clustering starting threshold; If clustering is successful, the CFD track is started, the confidence is set to the first set value, and the center point and cluster radius of successful clustering are recorded; If clustering is unsuccessful, determine whether there is a previous frame to pass the point cloud data to be processed. If so, cluster the previous frame and the current frame to be processed point cloud data. If clustering is successful, record the cluster center point and cluster radius, mark the track status as TBC, and set the confidence to the second set value; if not, predict the position of the point cloud in the radar coordinate system at the next moment according to the vehicle's motion state, and pass it to the next frame of point cloud data to be processed.

8. A stationary vehicle target recognition device, characterized in that: The stationary vehicle target recognition device includes a processor, a memory, and a stationary vehicle target recognition program stored in the memory and executable by the processor, wherein when the stationary vehicle target recognition program is executed by the processor, the steps of the stationary vehicle target recognition method as described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a stationary vehicle target recognition program, wherein when the stationary vehicle target recognition program is executed by a processor, the steps of the stationary vehicle target recognition method according to any one of claims 1 to 6 are implemented.

Images