Target detection method, device and equipment, and storage medium

Abstract

The application provides a target detection method, device and equipment. The target detection method comprises the following steps: when obtaining measured laser point cloud data reflected by a target, generating a head detection frame of a target vehicle according to the measured laser point cloud data, and generating a prior detection frame of the target on the head detection frame according to prior data of the target, wherein the prior data at least comprises length, height and width of the target, height information of a geometric center point of the target from the ground, and offset information of the geometric center point of the target from a geometric center point of the head; generating calibration data according to the measured laser point cloud data reflected by the target, adjusting the prior detection frame according to the calibration data, and obtaining a detection frame of the target. The application can solve the problem of how to detect other accessories (such as rearview mirrors) of the vehicle, thereby improving the accuracy of target vehicle detection during driving of the container truck and improving the safety factor of driving of the container truck.

Description

TECHNICAL FIELD

[0001] The present application relates to the target detection technology in the field of unmanned driving, and particularly relates to a target detection method, device and equipment, and a storage medium. BACKGROUND

[0002] With the wide application of laser radar technology in the field of unmanned driving, the container truck for unmanned driving based on laser radar detection technology is also widely applied to port operations. In the driving process of the container truck for unmanned driving, the surrounding environment information is acquired through various sensors such as laser radar, millimeter wave radar and camera, so as to realize the detection of various target vehicles. Whether the detection of the target vehicle is accurate determines whether the planning of the driving of the container truck is reasonable, and further determines whether the driving of the container truck is safe, so whether the detection of the target vehicle is accurate is very important.

[0003] At present, when the target vehicle is detected, only the vehicle head and vehicle body of other container trucks can be detected, and other accessories (such as rearview mirrors) of the vehicle cannot be detected, so that a reasonable safety distance cannot be reserved when planning the driving of the container truck. Reserving a reasonable safety distance can avoid the vehicle collision caused by too small vehicle distance, and also avoid the occupation of normal driving space caused by too large vehicle distance.

[0004] Therefore, how to detect other accessories (such as rearview mirrors) of the vehicle so as to improve the accuracy of the detection of the target vehicle in the driving process of the container truck and improve the safety factor of the driving of the container truck still needs to be considered. SUMMARY

[0005] The present application provides a target detection method, device and equipment, and a storage medium, to solve the problem of how to detect other accessories (such as rearview mirrors) of the vehicle so as to improve the accuracy of the detection of the target vehicle in the driving process of the container truck and improve the safety factor of the driving of the container truck.

[0006] In one aspect, the present application provides a target detection method, comprising:

[0007] When the measured laser point cloud data reflected by the target is acquired, a vehicle head detection frame of the target vehicle is generated according to the measured laser point cloud data, and a priori detection frame of the target on the vehicle head detection frame is generated according to priori data of the target, wherein the priori data at least includes length, height and width of the target, height information of a geometric center point of the target from the ground, and offset information of the geometric center point of the target from the geometric center point of the vehicle head.

[0008] The calibration data is generated according to the measured laser point cloud data reflected by the target object, and the prior detection frame is adjusted according to the calibration data to obtain a detection frame of the target object corresponding to the measured laser point cloud data, wherein the calibration data at least includes a calibration width of the target object and a calibration geometric center point of the target object.

[0009] When the prior detection frame is adjusted, the length and height of the target object, the height information of the geometric center point of the target object from the ground, the length direction of the target object consistent with the forward direction of the target vehicle, the width direction of the target object parallel to the ground, and the height direction of the target object perpendicular to the length direction and the width direction, respectively.

[0010] In one of the embodiments, the method further comprises:

[0011] When it is determined that the target object enters the blind area, the first N frames of laser point cloud frames corresponding to the laser point cloud frame when the target object enters the blind area are obtained, wherein one of the ways to determine that the target object enters the blind area is that when it is determined that only one of the plurality of line beams emitted by the laser radar can be reflected by the target object, it is determined that the target object enters the blind area.

[0012] The historical calibration data is generated when the plurality of prior detection frames corresponding to the first N frames of laser point cloud frames are adjusted based on the point cloud data of the target object, wherein the historical calibration data at least includes a plurality of calibration widths of the target object, a plurality of calibration geometric center points of the target object, and offset information between each of the plurality of calibration geometric center points and the geometric center point of the vehicle head.

[0013] The detection frame after the target object enters the blind area is generated according to the historical calibration data and the measurement information used when the plurality of prior detection frames are generated, wherein the measurement information at least includes the length and height of the target object, and the height information of the geometric center point of the target object from the ground.

[0014] In one of the embodiments, the detection frame after the target object enters the blind area is generated according to the historical calibration data and the measurement information used when the plurality of prior detection frames are generated, which comprises:

[0015] The average width determined according to the plurality of calibration widths in the historical calibration data, the average geometric center point determined according to the plurality of calibration geometric center points of the target object, and the average offset determined according to the offset information between each of the plurality of calibration geometric center points and the geometric center point of the vehicle head are obtained.

[0016] The detection frame after the target object enters the blind area is generated based on the average width, the average geometric center point, the average offset, and the measurement information.

[0017] In one embodiment, after generating a detection frame for the target object entering the blind spot, the method further includes:

[0018] When the number of point clouds determined according to the reflected beam of the target object is less than the first number, the detection frame after the target object enters the blind spot is used as the current detection frame of the target object;

[0019] When the number of point clouds determined based on the reflected beam of the target object is greater than or equal to the first number, it is determined that the target object is outside the blind spot, and the process returns to the step of obtaining the measured laser point cloud data reflected by the target object, generating a front detection frame of the target vehicle based on the measured laser point cloud data, and generating a priori detection frame of the target object on the front detection frame based on the prior data of the target object.

[0020] In one embodiment, generating calibration data based on the measured laser point cloud data reflected by the target object, adjusting the prior detection frame based on the calibration data, and obtaining a detection frame of the target object corresponding to the measured laser point cloud data includes:

[0021] Generate the point cloud gravity center based on the measured laser point cloud data reflected by the target object, and use the point cloud gravity center as the calibration geometric center point of the target object;

[0022] Updating the geometric center point of the target object in the prior detection frame to the calibration geometric center point to obtain an adjusted detection frame;

[0023] The calibrated width of the target is determined according to the measured laser point cloud data reflected by the target and the adjusted detection frame, and the width of the adjusted detection frame is updated according to the calibrated width of the target to obtain the detection frame of the target corresponding to the measured laser point cloud data.

[0024] In one embodiment, determining the calibration width of the target object based on the measured laser point cloud data reflected by the target object and adjusting the detection frame includes:

[0025] Determining an edge point cloud in the adjusted detection frame that is away from a geometric center point of the vehicle head based on the measured laser point cloud data reflected by the target object and the coverage of the adjusted detection frame;

[0026] According to the position information of the edge point cloud and the position information of the edge of the adjustment detection frame, when it is determined that the edge of the adjustment detection frame covers the edge point cloud, the width of the adjustment detection frame is the calibrated width of the target object, wherein the edge refers to the edge in the adjustment detection frame that is away from the front of the vehicle and extends in the high direction of the adjustment detection frame. When the edge covers the edge point cloud, the positions of other edges in the adjustment detection frame except the edge remain unchanged.

[0027] In one embodiment, the method further comprises:

[0028] When the number of laser point clouds covered by the detection box of the target object corresponding to the measured laser point cloud data is less than the second number, it is determined that the detection box of the target object corresponding to the measured laser point cloud data is an unusable detection box.

[0029] In one embodiment, the generating a priori detection box of the target object on the vehicle head detection box according to the priori data of the target object comprises:

[0030] generating an initial priori detection box of the target object on the vehicle head detection box based on the length, height and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information between the geometric center point of the target object and the geometric center point of the vehicle head;

[0031] increasing the length and width of the initial priori detection box within a preset range to obtain the priori detection box of the target object, wherein the extension direction of the initial priori detection box when the width is increased is away from the position of the target vehicle.

[0032] On the other hand, the present application provides a target object detection device, comprising:

[0033] an acquisition module configured to acquire measured laser point cloud data reflected by a target object;

[0034] a processing module configured to, when the measured laser point cloud data reflected by the target object is acquired, generate a vehicle head detection box of a target vehicle according to the measured laser point cloud data, and generate a priori detection box of the target object on the vehicle head detection box according to priori data of the target object, wherein the priori data at least includes the length, height and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information between the geometric center point of the target object and the geometric center point of the vehicle head;

[0035] an adjustment module configured to generate calibration data according to the measured laser point cloud data reflected by the target object, and adjust the priori detection box according to the calibration data to obtain a detection box of the target object corresponding to the measured laser point cloud data, wherein the calibration data at least includes the calibration width of the target object and the calibration geometric center point of the target object;

[0036] When the priori detection box is adjusted, the length and height of the target object, the height information of the geometric center point of the target object from the ground are all unchanged, the length direction of the target object is consistent with the forward direction of the target vehicle, the width direction of the target object is parallel to the ground, and the height direction of the target object is perpendicular to the length direction and the width direction, respectively.

[0037] On the other hand, the present application provides an electronic device, comprising a processor and a memory in communication connection with the processor;

[0038] the memory stores computer execution instructions;

[0039] The processor executes computer-executed instructions stored in the memory to implement the target object detection method according to the first aspect.

[0040] In another aspect, the present application provides a computer-readable storage medium, which stores computer-executed instructions, when the instructions are executed, causing a computer to execute the target object detection method according to the first aspect.

[0041] In another aspect, the present application provides a computer program product, which comprises a computer program, when the computer program is executed by a processor, implementing the target object detection method according to the first aspect.

[0042] The target object detection method provided by the embodiments of the present application collects prior data of the target object in advance, generates a prior detection frame of the target object on the head detection frame of the target vehicle after generating the head detection frame of the target vehicle in actual measurement according to the prior data. Then, calibration data is generated according to the measured laser point cloud data reflected by the target object in actual measurement, and the prior detection frame is adjusted according to the calibration data to obtain a detection frame of the target object corresponding to the measured laser point cloud data. In this way, the detection frame of the actually detected target object is obtained.

[0043] After the detection frame of the target object is obtained, a safety distance can be set when planning the driving of the container truck to prevent collision with the target object. Therefore, the method provided by the present application can realize detection of other accessories (such as rearview mirrors) of the vehicle, thereby improving the accuracy of detection of the target vehicle during the driving of the container truck and improving the safety factor of the driving of the container truck. In addition, the efficiency of actual detection of other accessories can also be improved by establishing the prior detection frame of the other accessories of the vehicle in advance.

[0044] In addition, the embodiments of the present application also illustrate how to maintain the detection frame of the target object when the target object enters the blind area, which can greatly increase the driving safety of the vehicle and improve the transportation efficiency of the container truck. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0046] Figure 1 An application scenario diagram of the target object detection method provided by the present application is shown in FIG. 1.

[0047] Figure 2 A flowchart of the target object detection method provided by an embodiment of the present application is shown in FIG. 2.

[0048] Figure 3A direction diagram of length, width and height of an object provided for an embodiment of the present application;

[0049] Figure 4 A width adjustment diagram of adjusting a detection frame provided for an embodiment of the present application;

[0050] Figure 5 Another flow diagram of a detection method of an object provided for an embodiment of the present application;

[0051] Figure 6 A diagram of a detection device of an object provided for an embodiment of the present application;

[0052] Figure 7 A diagram of an electronic device provided for an embodiment of the present application.

[0053] The specific embodiments of the present disclosure have been shown through the above-described drawings, and will be described in more detail hereinafter. The drawings and the written description are not intended to limit the scope of the present disclosure in any way, but to explain the concept of the present disclosure to those skilled in the art by referring to a specific embodiment. DETAILED DESCRIPTION

[0054] Exemplary embodiments will be described in detail herein below with reference to the drawings. In the following description, the same drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present disclosure. Instead, they only describe example devices and methods consistent with some aspects of the present disclosure, as detailed in the appended claims.

[0055] In the description of the present application, it should be understood that the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0056] With the wide application of laser radar technology in the field of unmanned driving, container trucks for unmanned driving based on laser radar detection technology are also widely used in port operations. During the driving of the unmanned container truck, various sensors such as laser radar, millimeter wave radar and camera are used to obtain surrounding environment information to realize the detection of various target vehicles. By detecting the target vehicle, the driving route of the container truck can be planned. Therefore, whether the detection of the target vehicle is accurate determines whether the planning of the container truck driving is reasonable, and further determines whether the container truck driving is safe.

[0057] However, the existing detection technology often only gives the detection frame of the vehicle head and body when detecting other container trucks, and ignores other vehicle accessories such as rearview mirrors. If no redundant safety distance is reserved in the planning process, safety accidents may occur during vehicle driving, causing economic losses. However, too large safety distance will also occupy normal driving space and affect production efficiency.

[0058] Therefore, the present application provides a target detection method, device and equipment, and a storage medium. The target is, for example, a rearview mirror. The target detection method comprises: when obtaining measured laser point cloud data reflected by the target, generating a vehicle head detection frame of a target vehicle according to the measured laser point cloud data, and generating a prior detection frame of the target on the vehicle head detection frame according to prior data of the target, the prior data being laser point cloud data generated when detecting the target before this detection; generating calibration data according to the measured laser point cloud data reflected by the target, adjusting the prior detection frame according to the calibration data to obtain a detection frame of the target corresponding to the measured laser point cloud data, wherein the calibration data at least includes a calibration width of the target and a calibration geometric center point of the target; when adjusting the prior detection frame, the length and height of the target, the height information of the geometric center point of the target from the ground, the length direction of the target consistent with the forward direction of the target vehicle, the width direction of the target parallel to the ground, and the height direction of the target perpendicular to the length direction and the width direction, respectively.

[0059] That is, the prior data of the target is collected in advance, and after the vehicle head detection frame of the target vehicle is generated in actual measurement, the prior detection frame of the target is generated on the vehicle head detection frame according to the prior data. Then, the calibration data is generated according to the measured laser point cloud data reflected by the target in actual measurement, and the prior detection frame is adjusted according to the calibration data to obtain the detection frame of the target corresponding to the measured laser point cloud data. In this way, the detection frame of the target actually detected is obtained. After obtaining the detection frame of the target, a safety distance can be set when planning the driving of the container truck to prevent collision with the target. Therefore, the method provided by the present application can realize the detection of other accessories (such as rearview mirrors) of the vehicle, thereby improving the accuracy of the detection of the target vehicle during the driving of the container truck and improving the safety factor of the driving of the container truck. In addition, the efficiency of the actual detection of other accessories can also be improved by collecting prior data in advance.

[0060] The data collection method provided by the present application is applied to an electronic device, such as a controller, a processor, etc. provided on a container truck for unmanned driving planning, or a controller, a processor, etc. in wireless remote communication with a laser radar device on a container truck. Figure 1An application diagram of the target detection method provided in the present application is shown in the figure, taking the rearview mirror on the container truck as the target. When the laser radar receives the reflected light beam of the target, the measured laser point cloud data is generated. The electronic device obtains the measured laser point cloud data of the target reflected by the laser radar, generates a head detection box of the target vehicle according to the measured laser point cloud data, and generates a prior detection box of the target on the head detection box according to the prior data of the target. The calibration data is generated according to the measured laser point cloud data of the target, and the prior detection box is adjusted according to the calibration data to obtain the detection box of the target corresponding to the measured laser point cloud data.

[0061] Please refer to Figure 2 An embodiment of the present application provides a target detection method, comprising:

[0062] S210, when obtaining the measured laser point cloud data of the target, generating a head detection box of the target vehicle according to the measured laser point cloud data, and generating a prior detection box of the target on the head detection box according to the prior data of the target. The prior data is the laser point cloud data generated when the target is detected before this detection.

[0063] Prior data of the target is collected before actual target detection. Specifically, the target (such as the rearview mirror) on the container truck is measured by the laser radar or other devices to determine the prior data of the target on the container truck. The prior data at least includes the length, height and width of the target, the height information of the geometric center point of the target from the ground, and the offset information (including distance and direction) of the geometric center point of the target from the geometric center point of the head. The length direction of the target is consistent with the forward direction of the target vehicle, the width direction of the target is parallel to the ground, and the height direction of the target is perpendicular to the length direction and the width direction respectively.

[0064] When obtaining the measured laser point cloud data of the target, the head detection box of the target vehicle is generated according to the measured laser point cloud data. The measured laser point cloud data is a frame of laser point cloud data, which can be considered as the Lth frame of laser point cloud data (L is a natural number greater than zero). After generating the head detection box, the prior detection box of the target is generated on the head detection box according to the prior data of the target.

[0065] Taking the rearview mirror as an example, according to these prior data, when the prior detection box of the target is generated on the head detection box according to the prior data of the target, the prior detection boxes of the left and right rearview mirrors are generated at the head of the container truck. As shown in Figure 3 The prior detection box is close to the head detection box, and the head direction of the prior detection box is consistent with the head by default. The length direction of the prior detection box is consistent with the forward direction of the target vehicle, as shown in Figure 3A direction of the middle mark. The width direction of the prior detection frame is parallel to the ground, as shown in Figure 3 B direction of the middle mark. The height direction of the prior detection frame is perpendicular to the length direction and the width direction, as shown in Figure 3 C direction of the middle mark.

[0066] In an optional embodiment, an initial prior detection frame can be established first, and then the length and width of the initial prior detection frame are expanded to obtain the prior detection frame. That is, based on the length, height and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information between the geometric center point of the target object and the geometric center point of the vehicle head, an initial prior detection frame of the target object is generated. The length and width of the initial prior detection frame are increased within a preset range to obtain the prior detection frame of the target object, wherein the extension direction of the width of the initial prior detection frame is away from the position of the target vehicle. The preset range can be set according to actual needs, and the embodiment is not limited.

[0067] When the width is expanded, only the vehicle outer side is expanded to prevent all the points in the prior detection frame after expansion from being in the vehicle head detection frame, so that the error of negative mirror width occurs in the adjustment process. After the length and width are expanded, the points contained in the prior detection frame obtained after expansion are added to the point cloud of the target object, so that the shape and position of the prior detection frame can be better adjusted according to the point cloud distribution in the subsequent process.

[0068] Because there may be deviations in the direction and position of the detected vehicle head detection frame and the actual vehicle head, the prior detection frame of the target object (such as the rearview mirror) cannot accurately describe the actual position information of the rearview mirror. Therefore, during actual detection, the prior detection frame needs to be adjusted according to the actually collected laser point cloud.

[0069] In an optional embodiment, the prior detection frame is adjusted only when it is determined according to the measured laser point cloud data reflected by the target object on the target vehicle that the distance between the target vehicle and the container truck is less than a preset distance. If the distance between the target vehicle and the container truck is greater than or equal to the preset distance, the adjustment operation of the prior detection frame is not needed, and the prior detection frame can be directly output as the detection frame of the target object. The distance between the target vehicle and the container truck can be the distance between the two vehicle heads, or the distance between the vehicle head of the target vehicle and the vehicle tail of the container truck. When the distance between the target vehicle and the container truck refers to the distance between the two vehicle heads, the preset distance can be greater than or equal to 30 meters and less than or equal to 45 meters. When the distance between the target vehicle and the container truck refers to the distance between the vehicle head of the target vehicle and the vehicle tail of the container truck, the preset distance can be greater than or equal to 15 meters and less than or equal to 30 meters.

[0070] S220, generating calibration data according to the measured laser point cloud data reflected by the target object, adjusting the prior detection frame according to the calibration data to obtain a detection frame of the target object corresponding to the measured laser point cloud data, wherein the calibration data at least includes a calibration width of the target object and a calibration geometric center point of the target object.

[0071] It should be emphasized that when adjusting the prior detection frame, the length and height of the target object, the height information of the geometric center point of the target object from the ground, the length direction of the target object consistent with the forward direction of the target vehicle, the width direction of the target object parallel to the ground, and the height direction of the target object perpendicular to the length direction and the width direction are all unchanged.

[0072] That is, when adjusting the prior detection frame, the front, back, left and right positions of the center point of the prior detection frame and the width of the prior detection frame are adjusted.

[0073] First, it is assumed that the height position of the prior detection frame is unchanged, that is, the height information of the geometric center point of the target object from the ground is unchanged. In addition, it is assumed that the length and height of the target object are unchanged. Then, a point cloud barycenter is generated according to the measured laser point cloud data reflected by the target object, and the point cloud barycenter is taken as the calibration geometric center point of the target object. After updating the geometric center point of the target object in the prior detection frame to the calibration geometric center point, an adjusted detection frame is obtained. That is, the point cloud barycenter in the prior detection frame is calculated, and then the position of the geometric center point of the target object in the prior detection frame is moved to the position of the point cloud barycenter to obtain the adjusted detection frame.

[0074] Then, the width of the adjusted detection frame is adjusted until the boundary of the adjusted detection frame coincides with the point farthest from the vehicle head. That is, the calibration width of the target object is determined according to the measured laser point cloud data reflected by the target object and the adjusted detection frame, and the width of the adjusted detection frame is updated according to the calibration width of the target object to obtain a detection frame of the target object corresponding to the measured laser point cloud data. Specifically, according to the measured laser point cloud data reflected by the target object and the coverage range of the adjusted detection frame, an edge point cloud of the geometric center point farthest from the vehicle head in the adjusted detection frame is determined. According to the position information of the edge point cloud and the position information of the edge of the adjusted detection frame, when the edge edge of the adjusted detection frame covers the edge point cloud, the width of the adjusted detection frame is the calibration width of the target object.

[0075] The edge edge refers to the edge of the adjusted detection frame farthest from the vehicle head and extending in the direction of the height of the adjusted detection frame. When the edge edge covers the edge point cloud, the positions of the other edges of the adjusted detection frame except the edge edge are unchanged. That is, when adjusting the width of the adjusted detection frame, the edge edge of the adjusted detection frame covers the edge point cloud, and the adjustment is completed. Figure 4 Figure 4 ​The a point identified in the middle is an edge point cloud. When adjusting the detection frame, the edge side (e.g. Figure 4 the b side identified in the middle) of the detection frame is adjusted to coincide with the edge point cloud, Figure 4 The dashed line in the middle is the adjusted edge side.

[0076] In an optional embodiment, during the adjustment of the prior detection frame, or after obtaining the detection frame of the target object corresponding to the measured laser point cloud data, it is necessary to determine whether the detection frame is available by judging the number of point clouds covered by the detection frame. When the number of laser point clouds covered by the detection frame of the target object corresponding to the measured laser point cloud data is less than a second number, it is determined that the detection frame of the target object corresponding to the measured laser point cloud data is an unavailable detection frame. The threshold range of the second number is, for example, greater than 0 and less than or equal to 10. The unavailable detection frame can be directly deleted, so as to prevent the point cloud noise from affecting the detection result.

[0077] After completing the adjustment of the geometric center point and the width of the prior detection frame, the detection frame of the target object corresponding to the measured laser point cloud data is obtained. Based on the detection frame of the target object, a route can be planned, and the planned route makes the subject container truck not collide with the target object (such as a rearview mirror) of the target vehicle (other container trucks) when driving. In this way, by detecting other accessories of the vehicle (such as a rearview mirror), the accuracy of detecting the target vehicle during the driving of the container truck is improved, and the safety factor of the driving of the container truck is improved.

[0078] During the driving of the actual container truck and the target vehicle, sometimes the required amount of laser point cloud data cannot be obtained to adjust the prior detection frame, for example, the target object enters a blind area or other situations. The inability to generate the detection frame of the target object will cause a certain degree of driving danger, so how to maximize the safety of the driving of the vehicle is crucial.

[0079] In an optional embodiment, when it is determined that only one line bundle of the multiple line bundles emitted by the lidar can be reflected by the target object, it is determined that the target object enters a blind area. When it is determined that the target object enters the blind area, the prior detection frame of the target object is generated according to the calibration data of the prior detection frame calibrated before the target object enters the blind area.

[0080] Specifically, when it is determined that the target object enters the blind area, the first N frames of laser point cloud frames corresponding to the laser point cloud frame when the target object enters the blind area are obtained (N is a natural number greater than zero, for example, N is equal to 3). The historical calibration data generated when the plurality of prior detection boxes corresponding to the first N frames of laser point cloud frames are adjusted based on the point cloud data of the target object of the first N frames of laser point cloud frames. The historical calibration data at least includes a plurality of calibration widths of the target object, a plurality of calibration geometric center points of the target object, and offset information (including distance and direction) between each calibration geometric center point in the plurality of calibration geometric center points and the geometric center point of the vehicle head. Further, the detection box after the target object enters the blind area is generated according to the historical calibration data and the measurement information used when the plurality of prior detection boxes are generated. The measurement information at least includes the length and height of the target object, and the height information of the geometric center point of the target object from the ground.

[0081] As described above, for any one frame of laser point cloud data reflected by the target object, when the prior detection box corresponding to the any one frame is adjusted, calibration data is generated, and the prior detection box is adjusted based on the calibration data. The calibration data at least includes the calibration width of the target object and the calibration geometric center point of the target object. When the plurality of prior detection boxes corresponding to the first N frames of laser point cloud frames are adjusted based on the point cloud data of the target object of the first N frames of laser point cloud frames, the historical calibration data generated is at least includes a plurality of calibration widths of the target object and a plurality of calibration geometric center points of the target object. It should be noted that the historical calibration data further includes offset information (including distance and direction) between each calibration geometric center point in the plurality of calibration geometric center points and the geometric center point of the vehicle head.

[0082] When generating the detection box of the target object in the blind area, only the historical calibration data is not enough, and the length and height of the target object, the height information of the geometric center point of the target object from the ground, etc. are also needed. That is, the measurement information used when the plurality of prior detection boxes are generated is obtained. Based on the historical calibration data and the measurement information, the detection box after the target object enters the blind area can be generated.

[0083] More specifically, the average width determined based on multiple calibration widths in the historical calibration data, the average geometric center point determined based on multiple calibration geometric center points of the target, and the average offset determined based on the offset information between each calibration geometric center point in the multiple calibration geometric center points and the geometric center point of the vehicle head are obtained. Based on the average width, the average geometric center point, the average offset, and the measurement information used, a detection frame is generated after the target enters the blind spot. The average width, the average geometric center point, and the average offset are all obtained by summing and averaging. When determining the average geometric center point, the position information of the average geometric center point is summed and averaged based on the position information of the multiple calibration geometric center points, thereby determining the average geometric center point. Because the offset includes distance and direction, when determining the average offset, the average distance is obtained by summing and averaging multiple distances, and then the directions are averaged to obtain the average direction.

[0084] After generating a detection frame for the target object after it enters the blind spot, if the target object is still within the blind spot, the generated detection frame for the target object after it enters the blind spot is still used as the current detection frame for the target object. For example, if the number of point clouds determined based on the reflected beam of the target object is less than the first number, it is determined that the target object is still within the blind spot, and the detection frame for the target object after it enters the blind spot is used as the current detection frame for the target object.

[0085] If the number of point clouds determined based on the reflected beam of the target object is greater than or equal to the first number, the target object is determined to be outside the blind spot, and the process returns to step S210. That is, if the target object is determined to be outside the blind spot, the process for generating a detection frame for the target object when the target object is outside the blind spot, as described above, is performed. Specifically, the prior detection frame is adjusted based on the laser point cloud data reflected by the target object, ultimately generating a detection frame for the target object.

[0086] For ease of understanding, combined Figure 5 The specific process of the target object detection method is further described. Figure 5 As shown, multiple prior detection frames for the target object are first generated based on the measurement information. The system then determines whether the distance between the target vehicle and the ego vehicle (this container truck) is less than a preset distance. If the distance between the target vehicle and the ego vehicle (this container truck) is less than the preset distance, calibration data is generated based on the measured laser point cloud data reflected by the target object, and the prior detection frames are adjusted based on this calibration data. If the distance between the target vehicle and the ego vehicle (this container truck) is greater than or equal to the preset distance, the prior detection frame is directly output as the detection frame for the target object.

[0087] In the real-time target detection process, it is also necessary to determine whether the target is in the blind area of the laser radar (i.e., whether the target enters the blind area), and if the target is in the blind area, a detection frame of the target entering the blind area needs to be generated according to historical calibration data and measurement information. If the target is not in the blind area, the adjusted prior detection frame of the target is output as the detection frame of the target.

[0088] In summary, the target detection method provided in the embodiment collects prior data of the target in advance, generates a prior detection frame of the target on the head detection frame of the target vehicle according to the prior data after generating the head detection frame of the target vehicle in actual measurement. Then, calibration data is generated according to the measured laser point cloud data reflected by the target in actual measurement, and the prior detection frame is adjusted according to the calibration data to obtain a detection frame of the target corresponding to the measured laser point cloud data. In this way, the detection frame of the actually detected target is obtained.

[0089] After obtaining the detection frame of the target, a safety distance can be set when planning the driving of the container truck to prevent collision with the target. Therefore, the method provided in the application can realize detection of other accessories (such as rearview mirrors) of the vehicle, thereby improving the accuracy of detection of the target vehicle during driving of the container truck and improving the safety factor of driving of the container truck. In addition, the efficiency of actual detection of other accessories can be improved by establishing a prior detection frame of the other accessories of the vehicle in advance.

[0090] In addition, it is also illustrated how to maintain the detection frame of the target when the target enters the blind area, which can greatly increase the driving safety of the vehicle and improve the transportation efficiency of the container truck.

[0091] See Figure 6 One embodiment of the application also provides a target detection device 10, comprising:

[0092] The acquisition module 11 acquires measured laser point cloud data reflected by the target;

[0093] The processing module 12 is configured to generate a head detection frame of the target vehicle according to the measured laser point cloud data when the measured laser point cloud data reflected by the target is acquired, and generate a prior detection frame of the target on the head detection frame of the target vehicle according to prior data of the target. The prior data at least includes the length, height and width of the target, the height information of the geometric center point of the target from the ground, and the offset information of the geometric center point of the target from the geometric center point of the head.

[0094] The adjusting module 13 is configured to generate calibration data according to the measured laser point cloud data reflected by the target object, and adjust the prior detection frame according to the calibration data to obtain a detection frame of the target object corresponding to the measured laser point cloud data, wherein the calibration data at least includes a calibration width of the target object and a calibration geometric center point of the target object. When the prior detection frame is adjusted, the length and height of the target object, the height information of the geometric center point of the target object from the ground, the length direction of the target object consistent with the forward direction of the target vehicle, the width direction of the target object parallel to the ground, and the height direction of the target object perpendicular to the length direction and the width direction are all unchanged.

[0095] The target object detection device 10 further includes a detection frame generation module 14. When it is determined that the target object enters the blind area, the detection frame generation module 14 is configured to obtain the first N frames of laser point cloud frames corresponding to the laser point cloud frame when the target object enters the blind area, wherein one of the ways to determine that the target object enters the blind area is that only one of the multiple beams emitted by the laser radar can be reflected by the target object. The detection frame generation module 14 is further configured to obtain historical calibration data generated when adjusting the multiple prior detection frames of the target object based on the point cloud data of the target object based on the first N frames of laser point cloud frames, wherein the historical calibration data at least includes multiple calibration widths of the target object, multiple calibration geometric center points of the target object, and offset information between each of the multiple calibration geometric center points and the geometric center point of the vehicle head. The detection frame generation module 14 is further configured to generate a detection frame of the target object after entering the blind area according to the historical calibration data and the measurement information used when generating the multiple prior detection frames, wherein the measurement information at least includes the length and height of the target object, and the height information of the geometric center point of the target object from the ground.

[0096] The detection frame generation module 14 is specifically configured to obtain a width average value determined according to the multiple calibration widths in the historical calibration data, an average geometric center point determined according to the multiple calibration geometric center points of the target object, and an average offset determined according to the offset information between each of the multiple calibration geometric center points and the geometric center point of the vehicle head. The detection frame generation module 14 is further configured to generate a detection frame of the target object after entering the blind area based on the width average value, the average geometric center point, the average offset, and the measurement information used.

[0097] The detection frame generation module 14 is further configured to, when the number of point clouds determined according to the reflected beam of the target object is less than the first number, take the detection frame of the target object after entering the blind area as the current detection frame of the target object; and when the number of point clouds determined according to the reflected beam of the target object is greater than or equal to the first number, determine that the target object is outside the blind area, and return to execute the step of matching the prior detection frame corresponding to the Lth historical laser point cloud data from the multiple prior detection frames as the prior detection frame when the measured laser point cloud data reflected by the target object is obtained.

[0098] The adjustment module 13 is specifically configured to generate a point cloud barycenter according to the measured laser point cloud data reflected by the target object, take a calibration geometric center point of the target object as the point cloud barycenter, and obtain an adjusted detection frame by updating the geometric center point of the target object in the prior detection frame to the calibration geometric center point. The adjustment module 13 is specifically configured to determine a calibration width of the target object according to the measured laser point cloud data reflected by the target object and the adjusted detection frame, and update the width of the adjusted detection frame according to the calibration width of the target object to obtain the detection frame of the target object corresponding to the measured laser point cloud data.

[0099] The adjustment module 13 is specifically configured to determine an edge point cloud of the geometric center point away from the vehicle head in the adjusted detection frame according to the measured laser point cloud data reflected by the target object and the coverage range of the adjusted detection frame. The adjustment module 13 is specifically configured to determine that the width of the adjusted detection frame is the calibration width of the target object when an edge edge of the adjusted detection frame covers the edge point cloud according to the position information of the edge point cloud and the position information of the edge of the adjusted detection frame, wherein the edge edge refers to an edge of the adjusted detection frame away from the vehicle head and extending in the direction of the height of the adjusted detection frame, and the position of the other edges of the adjusted detection frame except the edge edge is unchanged when the edge edge covers the edge point cloud.

[0100] The adjustment module 13 is further configured to determine that the detection frame of the target object corresponding to the measured laser point cloud data is an unusable detection frame when the number of laser point clouds covered by the detection frame of the target object corresponding to the measured laser point cloud data is less than the second number.

[0101] The processing module 12 is specifically configured to generate an initial prior detection frame of the target object on the vehicle head detection frame based on the length, height and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information between the geometric center point of the target object and the geometric center point of the vehicle head, and increase the length and width of the initial prior detection frame in a preset range to obtain the prior detection frame of the target object, wherein the extension direction of the initial prior detection frame when the width of the initial prior detection frame is increased is away from the position of the target vehicle.

[0102] See Figure 7 One embodiment of the present application also provides an electronic device 20, comprising a processor 21 and a memory 22 connected with the processor. The memory 22 stores computer execution instructions, and the processor 21 executes the computer execution instructions stored in the memory 22 to implement the target object detection method provided in any one of the above embodiments.

[0103] One embodiment of the present application also provides a computer readable storage medium, which stores computer execution instructions, and when the instructions are executed, the computer executes the target object detection method provided in any one of the above embodiments.

[0104] One embodiment of the present application further provides a computer program product comprising a computer program which, when executed by a processor, implements the target object detection method according to any one of the above embodiments.

[0105] It should be noted that the computer readable storage medium described above can be a Read Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Ferromagnetic Random Access Memory (FRAM), a Flash Memory, a magnetic surface memory, an optical disc, a Compact Disc Read-Only Memory (CD-ROM), or the like. It can also be various electronic devices including one or any combination of the above memories, such as a mobile phone, a computer, a tablet device, a personal digital assistant, and the like.

[0106] It should be noted that in this document, the term "comprising" or "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that processes, methods, articles or devices including a series of elements not only include those elements, but also include other elements not explicitly listed, or further include elements inherent in such processes, methods, articles or devices. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0107] The above-mentioned sequence numbers of embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0108] Through the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and a necessary general hardware platform, and of course, can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, or an optical disk) and includes a plurality of instructions for causing a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the methods described in the various embodiments of the present application.

[0109] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices generate a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0110] These computer program instructions can also be stored in a computer-readable memory that can cause the computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0112] The above is only the preferred embodiments of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent flow transformation made by using the contents of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method of detecting a target object, characterized by, The method comprises the following steps: When the measured laser point cloud data reflected by the target object is acquired, a head detection frame of the target vehicle is generated according to the measured laser point cloud data, and an initial prior detection frame of the target object is generated on the head detection frame according to prior data of the target object, the length and width of the initial prior detection frame are increased within a preset range to obtain a prior detection frame of the target object, wherein the extension direction of the width of the initial prior detection frame is away from the position of the target vehicle, and the prior data at least includes the length, height and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information of the geometric center point of the target object from the geometric center point of the head. A point cloud barycenter is generated according to the measured laser point cloud data reflected by the target object, and the calibration geometric center point of the target object is taken as the point cloud barycenter; the geometric center point of the target object in the prior detection frame is updated to the calibration geometric center point to obtain an adjusted detection frame; the calibration width of the target object is determined according to the measured laser point cloud data reflected by the target object and the adjusted detection frame, the width of the adjusted detection frame is updated according to the calibration width of the target object to obtain a detection frame of the target object corresponding to the measured laser point cloud data; wherein the measured laser point cloud data is a frame of laser point cloud data. When the prior detection frame is adjusted, the length and height of the target object, the height information of the geometric center point of the target object from the ground are all unchanged, the length direction of the target object is consistent with the forward direction of the target vehicle, the width direction of the target object is parallel to the ground, and the height direction of the target object is perpendicular to the length direction and the width direction respectively.

2. The method of claim 1, wherein, The method further comprises the following steps: When it is determined that the target object enters the blind area, the first N frames of laser point cloud frames corresponding to the laser point cloud frame in which it is determined that the target object enters the blind area are acquired, wherein one of the ways to determine that the target object enters the blind area is that only one of the multiple line beams emitted by the laser radar can be reflected by the target object. When the multiple prior detection frames corresponding to the first N frames of laser point cloud frames are adjusted based on the point cloud data of the target object, the historical calibration data generated are acquired, wherein the historical calibration data at least include multiple calibration widths of the target object, multiple calibration geometric center points of the target object, and offset information between each calibration geometric center point in the multiple calibration geometric center points and the geometric center point of the head. The detection frame after the target object enters the blind area is generated according to the historical calibration data and the measurement information used when the multiple prior detection frames are generated, wherein the measurement information at least includes the length and height of the target object, the height information of the geometric center point of the target object from the ground.

3. The method of claim 2, wherein, The detection frame after the target object enters the blind area is generated according to the historical calibration data and the measurement information used when the multiple prior detection frames are generated, which comprises the following steps: The average width determined according to the multiple calibration widths in the historical calibration data, the average geometric center point determined according to the multiple calibration geometric center points of the target object, and the average offset determined according to the offset information between each calibration geometric center point in the multiple calibration geometric center points and the geometric center point of the head are acquired. Generate a detection frame of the target object after entering the blind area based on the average width, the average geometric center point, the average offset, and the used measurement information.

4. The method of claim 2, wherein, After the detection frame of the target object after entering the blind area is generated, the method further includes: When the number of point clouds determined according to the reflection beam of the target object is less than a first number, taking the detection frame of the target object after entering the blind area as the current detection frame of the target object; When the number of point clouds determined according to the reflection beam of the target object is greater than or equal to the first number, determining that the target object is outside the blind area, and returning to execute the step of generating the detection frame of the target vehicle according to the measured laser point cloud data reflected by the target object, and generating the initial prior detection frame of the target object on the detection frame of the target vehicle according to the prior data of the target object.

5. The method of claim 1, wherein, The step of determining the calibrated width of the target object according to the measured laser point cloud data reflected by the target object and the adjusted detection frame includes: Determining an edge point cloud of the geometric center point of the adjusted detection frame away from the vehicle head according to the measured laser point cloud data reflected by the target object and the coverage range of the adjusted detection frame; When an edge edge of the adjusted detection frame covers the edge point cloud according to the position information of the edge point cloud and the position information of the edge of the adjusted detection frame, the width of the adjusted detection frame is the calibrated width of the target object, wherein the edge edge refers to an edge of the adjusted detection frame away from the vehicle head and extending in the direction of the height of the adjusted detection frame, and when the edge edge covers the edge point cloud, the positions of the other edges of the adjusted detection frame except the edge edge remain unchanged.

6. The method of claim 1, wherein, The method further includes: When the number of laser point clouds covered by the detection frame of the target object corresponding to the measured laser point cloud data is less than a second number, determining that the detection frame of the target object corresponding to the measured laser point cloud data is an unusable detection frame.

7. A device for detecting a target, characterized by The method includes: The acquisition module is configured to acquire measured laser point cloud data reflected by a target object. The processing module is configured to, when the measured laser point cloud data reflected by the target object is acquired, generate a detection frame of a target vehicle according to the measured laser point cloud data, and generate an initial prior detection frame of the target object on the detection frame of the target vehicle according to prior data of the target object, and increase the length and width of the initial prior detection frame within a preset range to obtain a prior detection frame of the target object, wherein the extension direction of the initial prior detection frame when the width of the initial prior detection frame is increased is away from the position of the target vehicle, and the prior data at least includes the length, height, and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information of the geometric center point of the target object from the geometric center point of the vehicle head. The processing module is configured to, when the measured laser point cloud data reflected by the target object is acquired, generate a detection frame of a target vehicle according to the measured laser point cloud data, and generate an initial prior detection frame of the target object on the detection frame of the target vehicle according to prior data of the target object, and increase the length and width of the initial prior detection frame within a preset range to obtain a prior detection frame of the target object, wherein the extension direction of the initial prior detection frame when the width of the initial prior detection frame is increased is away from the position of the target vehicle, and the prior data at least includes the length, height, and width of the target object, the height information of the geometric center point of the target object from the ground, and the offset information of the geometric center point of the target object from the geometric center point of the vehicle head. The adjusting module is configured to generate a point cloud barycenter according to the measured laser point cloud data reflected by the target object, take a calibration geometric center point of the target object as the point cloud barycenter, update a geometric center point of the target object in the prior detection frame to obtain an adjusted detection frame after taking the calibration geometric center point, determine a calibration width of the target object according to the measured laser point cloud data reflected by the target object and the adjusted detection frame, update a width of the adjusted detection frame according to the calibration width of the target object, and obtain a detection frame of the target object corresponding to the measured laser point cloud data; and the measured laser point cloud data is a frame of laser point cloud data. When the prior detection frame is adjusted, the length and height of the target object, the height information of the geometric center point of the target object from the ground, the length direction of the target object, the advancing direction of the target vehicle, the width direction of the target object, the height direction of the target object, and the length direction and the width direction of the target object are all unchanged.

8. An electronic device, comprising: The method comprises the following steps: a processor and a memory connected with the processor in communication; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to realize the target object detection method in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer execution instructions, and when the instructions are executed, the computer executes the target object detection method in any one of claims 1 to 6.

Images