Acquisition method and device of occupied data and storage medium

Abstract

The invention discloses an occupancy data acquisition method and device and a storage medium, and belongs to the technical field of vehicle control. The method comprises the following steps: acquiring point cloud information of continuous frames received by a radar point cloud sensor; identifying a bounding box of a dynamic target and a TrackID of the dynamic target contained in the information of the point cloud of the continuous frames; adjusting the point clouds contained in the bounding box of each dynamic target to obtain the adjusted point clouds of the dynamic targets; splicing the adjusted point clouds of the dynamic target to obtain the spliced point clouds of the dynamic target; obtaining the coordinates of the point cloud of the static target in the global coordinate system; splicing the point clouds of the static target under the global coordinate system to obtain the spliced point clouds of the static target; and combining the spliced point cloud of the dynamic target with the spliced point cloud of the static target to generate occupancy data of the road where the vehicle is located. The precision of the generated occupancy data of the road where the vehicle is located is improved while the cost is controlled.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of vehicle control, and particularly to a method, device, and storage medium for obtaining occupancy data. Background Art

[0002] With the application of lidar in vehicle driving, the method of judging the vehicle's feasible area through 3D (three-dimensional) BEV (Bird's-Eye View) occupancy data has also been applied in autonomous driving. Among them, occupancy data refers to data used to represent the presence or absence of objects in a specific area. In related technologies, on the one hand, 3D BEV occupancy data is obtained through visual recognition methods; on the other hand, 3D BEV occupancy data can be generated based on the labels of 3D lidar detection tasks or 3D segmentation tasks, combined with temporal information and Pose information.

[0003] In related technologies, the visual recognition method has the problem of insufficient accuracy in the generated occupancy data due to the lack of depth information in the visual ground truth system; the method of using the labels of 3D lidar detection tasks or 3D segmentation tasks has the problem of high cost due to the high price of 3D ground truth annotation. Therefore, how to control the data acquisition cost while ensuring the accuracy of occupancy data is a problem to be solved. Summary of the Invention

[0004] The embodiments of the present application provide a method, device, and storage medium for obtaining occupancy data, which can be used to control the data acquisition cost while ensuring the accuracy of occupancy data. The technical solutions are as follows:

[0005] On the one hand, the embodiments of the present application provide a method for obtaining occupancy data, the method includes:

[0006] Obtain the information of the point clouds of consecutive frames received by the radar point cloud sensor, and the point clouds of the consecutive frames are composed of the points where the laser pulses emitted by the radar point cloud sensor are reflected;

[0007] Identify the bounding boxes of the dynamic targets and the TrackID (unique identifier) of the dynamic targets included in the information of the point clouds of the consecutive frames;

[0008] Adjust the point clouds included in the bounding box of each dynamic target respectively to obtain the adjusted point clouds of the dynamic targets;

[0009] Stitch the adjusted point clouds of the dynamic targets to obtain the stitched point clouds of the dynamic targets;

[0010] Obtain the coordinates of the point cloud of the static target in the global coordinate system, where the point cloud of the static target is the remaining point cloud in the point cloud of the continuous frames except for the point cloud included in the bounding box of the dynamic target;

[0011] Stitch the point cloud of the static target in the global coordinate system to obtain the stitched point cloud of the static target;

[0012] Combine the stitched point cloud of the dynamic target with the stitched point cloud of the static target to generate the occupancy data of the road where the vehicle is located.

[0013] On the other hand, there is provided an apparatus for obtaining occupancy data, and the apparatus includes:

[0014] A first acquisition module, configured to acquire information of the point cloud of continuous frames received by a radar point cloud sensor, where the point cloud of the continuous frames is composed of points reflected by laser pulses emitted by the radar point cloud sensor;

[0015] An identification module, configured to identify the bounding box of the dynamic target and the TrackID of the dynamic target included in the information of the point cloud of the continuous frames;

[0016] An adjustment module, configured to respectively adjust the point cloud included in the bounding box of each dynamic target to obtain the adjusted point cloud of the dynamic target;

[0017] A first stitching module, configured to stitch the adjusted point cloud of the dynamic target to obtain the stitched point cloud of the dynamic target;

[0018] A second acquisition module, configured to acquire the coordinates of the point cloud of the static target in the global coordinate system, where the point cloud of the static target is the remaining point cloud in the point cloud of the continuous frames except for the point cloud included in the bounding box of the dynamic target;

[0019] A second stitching module, configured to stitch the point cloud of the static target in the global coordinate system to obtain the stitched point cloud of the static target;

[0020] A combination module, configured to combine the stitched point cloud of the dynamic target with the stitched point cloud of the static target to generate the occupancy data of the road where the vehicle is located.

[0021] On the other hand, there is also provided a non - temporary computer - readable storage medium, characterized in that a computer program is stored in the computer - readable storage medium, and the computer program is loaded and executed by a processor to implement the occupancy data acquisition method described in any one of the above.

[0022] On the other hand, a computer program product is also provided. The computer program product includes computer instructions which, when executed by a processor, implement the steps of the method for obtaining occupancy data described in any one of the above.

[0023] The technical solution provided by this application at least brings the following beneficial effects:

[0024] By obtaining the information of the point clouds of consecutive frames received by the radar point cloud sensor, identifying the bounding box information of the dynamic target and the TrackID of the dynamic target, adjusting the point clouds included in the bounding box of each dynamic target respectively, and splicing the point clouds of the adjusted dynamic targets, the spliced point clouds of the dynamic targets are obtained; by obtaining the coordinates of the point clouds of the static target in the global coordinate system and splicing the point clouds of the static target, the spliced point clouds of the static target are obtained; combining the spliced point clouds of the dynamic targets with the spliced point clouds of the static target to generate the occupancy data of the road where the vehicle is located, automatic annotation of dynamic targets and static targets is realized, while controlling the cost, the accuracy of the generated occupancy data of the road where the vehicle is located is improved, thereby reducing the risk of collision during vehicle driving and ensuring the safety of vehicle driving. Description of the Drawings

[0025] To more clearly illustrate the technical solutions in the embodiments of this application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0026] Figure 1 is a schematic diagram of an implementation environment provided by an embodiment of this application;

[0027] Figure 2 is a flowchart of a method for obtaining occupancy data provided by an embodiment of this application;

[0028] Figure 3 is a schematic structural diagram of an apparatus for obtaining occupancy data provided by an embodiment of this application. Detailed Embodiments

[0029] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe the embodiments of this application in detail with reference to the drawings.

[0030] An embodiment of this application provides a method for obtaining occupancy data. Please refer to Figure 1, which shows a schematic diagram of the method implementation environment provided by the embodiments of the present application. The implementation environment may include: HCU (Hybrid Control Unit, vehicle controller) 11, radar point cloud sensor 12, in-vehicle display screen 13, and camera 14.

[0031] Optionally, the radar point cloud sensors 12 are installed around the vehicle, used to emit laser pulses, receive the reflected signals corresponding to the laser pulses, and collect the information of the reflected signals and send it to the HCU 11. The cameras 14 are installed around the vehicle, used to capture video images of the road where the vehicle is located and send them to the HCU 11. The in-vehicle display screen 13 is installed in front of the driver's seat on the side, used to receive from the HCU 11 the video images of the road where the vehicle is located with the grids being occupied or not distinguished by different colors and display them.

[0032] Among them, the HCU 11, the radar point cloud sensors 12, and the in-vehicle display screen 13 establish a communication connection through a wired or wireless network.

[0033] Based on the above Figure 1 shown implementation environment, the embodiments of the present application provide a method for obtaining occupancy data as Figure 2 shown. Taking the method applied to the HCU (Hybrid Control Unit, vehicle controller) as an example, the method includes steps 201 - step 207.

[0034] In step 201, the HCU obtains the information of the point cloud of consecutive frames received by the radar point cloud sensor, and the point cloud of consecutive frames is composed of the points where the laser pulses emitted by the radar point cloud sensor are reflected.

[0035] In a possible implementation manner, the HCU controls the radar point cloud sensor to emit laser pulses. When the laser pulses encounter obstacles, they will be reflected and generate reflected signals. The HCU obtains the information of the point cloud of consecutive frames received by the radar point cloud sensor. Among them, the point cloud of consecutive frames is composed of the points where the laser pulses emitted by the radar point cloud sensor are reflected. The obstacles can be vehicles, pedestrians, animals, or objects, and the radar point cloud sensors can be installed around the vehicle.

[0036] Optionally, the HCU collects the information of the point cloud of consecutive frames received by the radar point cloud sensor, including: the HCU continuously collects the information of the reflected signals received by the radar point cloud sensor, and generates the information of the point cloud of consecutive frames based on the information of the reflected signals. Exemplarily, the information of the reflected signals includes the propagation direction of the reflected signals and the duration from the emission of the laser pulses by the radar point cloud sensor to the reception of the reflected signals, and the information of the point cloud of consecutive frames includes the positions of the reflection points, where the reflection points are the points where the laser pulses are reflected.

[0037] In a possible implementation, information of consecutive frames of point clouds is generated based on information of reflected signals, including: calculating the distance between a reflection point and a radar point cloud sensor based on the time duration from the emission of a laser pulse by the radar point cloud sensor to the reception of the reflected signal; determining the position of the reflection point based on the propagation direction of the reflected signal and the distance between the reflection point and the radar point cloud sensor.

[0038] Optionally, the position of the reflection point can be represented by the spatial coordinates of the reflection point in the spatial coordinate system of the radar point cloud sensor. Among them, the spatial coordinate system of the radar point cloud sensor can be established in advance. For example, the position where the radar point cloud sensor is located is used as the origin, the direction perpendicular to the road surface is used as the Z-axis, the direction perpendicular to the vehicle bumper is used as the Y-axis, and the direction parallel to the bumper is used as the X-axis.

[0039] In step 202, the HCU identifies the bounding box of the dynamic object and the TrackID of the dynamic object included in the information of consecutive frames of point clouds.

[0040] Exemplarily, after obtaining the information of consecutive frames of point clouds, the HCU identifies the bounding box of the dynamic object and the TrackID of the dynamic object included in the information of consecutive frames of point clouds, including: detecting the information of consecutive frames of point clouds, and identifying the dynamic objects included in each frame and the initial bounding box of the dynamic objects; tracking the movement trajectories of the same dynamic object in each frame, and establishing the TrackID of each dynamic object; corresponding the TrackID to the dynamic objects included in each frame; and correcting the initial bounding box of the dynamic objects included in each frame based on a preset timing model to obtain the final bounding box.

[0041] Optionally, the manner in which the HCU detects the information of consecutive frames of point clouds and identifies the dynamic objects included in each frame and the initial bounding box of the dynamic objects includes, but is not limited to: the HCU processes the information of each frame of point clouds through an object recognition algorithm to identify the dynamic objects, and segments the point clouds included in different dynamic objects through a clustering algorithm. The object recognition algorithm can also generate the initial bounding box of each dynamic object, and the HCU can obtain the position, size, and category of the initial bounding box through the object recognition algorithm.

[0042] In a possible implementation, after segmenting the point clouds included in different dynamic objects, the coordinates of the center point of each dynamic object and the length, width, and height of the dynamic object can be calculated based on the information of the point clouds included in each dynamic object.

[0043] Exemplarily, after detecting the information of the point cloud of consecutive frames, the HCU tracks the motion trajectory of the same dynamic target in each frame through a tracking algorithm, associates the same dynamic target in adjacent frames, and establishes a TrackID for each dynamic target. Among them, the tracking algorithm can be the Kalman filter method, the optical flow method, or the deep learning method. In a possible implementation, after determining the TrackID of each dynamic target, the HCU corresponds the TrackID to the dynamic target included in each frame, so as to identify the state of each dynamic target in different frames.

[0044] Optionally, the HCU further corrects the initial bounding box of the dynamic target included in each frame based on a preset timing model to obtain a final bounding box. Among them, the preset timing model is used to further analyze the motion trend of the dynamic target in consecutive frames according to the time sequence. The timing model can be LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Unit).

[0045] In a possible implementation, the HCU corrects the initial bounding box of the dynamic target included in each frame based on a preset timing model to obtain a final bounding box, including: the HCU identifies the motion trend of the dynamic target in consecutive frames based on the timing model, and adjusts the size and position of the initial bounding box of the dynamic target according to the motion trend in consecutive frames to obtain a final bounding box.

[0046] In step 203, the HCU adjusts the point cloud included in the bounding box of each dynamic target respectively to obtain the adjusted point cloud of the dynamic target.

[0047] Exemplarily, after completing the identification of the bounding box and TrackID of the dynamic target, the HCU adjusts the point cloud included in the bounding box of each dynamic target respectively, including: adjusting the position of the point cloud included in the bounding box of each dynamic target so that the center point of the dynamic target is translated to the coordinate origin, and rotating the orientation of the point cloud to a preset standard direction, where the coordinate origin is the origin of the coordinate system of the radar point cloud sensor.

[0048] Optionally, the point cloud included in the bounding box of the dynamic target is the point cloud of the dynamic target. Before adjusting the position of the point cloud of each dynamic target, the HCU calculates the spatial coordinates of the center point of the dynamic target based on the spatial coordinates of the point cloud of each dynamic target, and then subtracts the spatial coordinates of the center point of the dynamic target from the spatial coordinates of the point cloud of the dynamic target to realize translating the center point of the dynamic target to the coordinate origin.

[0049] In a possible implementation, after translating the point cloud of the dynamic target, the HCU rotates the orientation of the point cloud to a preset standard direction, including: calculating the rotation angle based on the current orientation of the dynamic target and the preset standard direction, constructing a rotation matrix according to the rotation angle and the rotation axis, and rotating the translated point cloud through the rotation matrix so that the orientation of the point cloud rotates to the preset standard direction, obtaining the adjusted point cloud of the dynamic target. Exemplarily, the preset standard direction can be set according to experience, and the Z-axis can be selected as the rotation axis.

[0050] In step 204, the HCU stitches the adjusted point cloud of the dynamic target to obtain the stitched point cloud of the dynamic target.

[0051] Exemplarily, after adjusting the point cloud of the dynamic target, the HCU stitches the adjusted point cloud of the dynamic target to obtain the stitched point cloud of the dynamic target, including: the HCU stitches the adjusted point cloud of the dynamic target through a stitching algorithm to obtain the stitched point cloud of the dynamic target, as the position of the dynamic target after adjustment.

[0052] In step 205, the HCU obtains the coordinates of the point cloud of the static target in the global coordinate system, where the point cloud of the static target is the remaining point cloud in the continuous frame of point clouds except for the point cloud included in the bounding box of the dynamic target.

[0053] In a possible implementation, after stitching the point cloud of the dynamic target, the remaining point cloud in the continuous frame of point clouds except for the point cloud included in the bounding box of the dynamic target is used as the point cloud of the static target, and the HCU obtains the coordinates of the point cloud of the static target in the global coordinate system, including: obtaining the coordinates of the point cloud of the static target in the coordinate system of the radar point cloud sensor; obtaining the position and orientation of the radar point cloud sensor in the global coordinate system; and converting the coordinates of the point cloud of the static target in the coordinate system of the radar point cloud sensor to the coordinates of the point cloud of the static target in the global coordinate system based on the position and orientation.

[0054] Optionally, the HCU generates the coordinates of the point cloud of the static target in the coordinate system of the radar point cloud sensor according to the information of the point cloud of the static target, and then obtains the position and orientation of the radar point cloud sensor in the global coordinate system through the vehicle navigation device. Among them, the global coordinate system can be established in advance. For example, the position of the vehicle is used as the origin, the direction perpendicular to the road surface is the A-axis, the direction perpendicular to the vehicle bumper is the B-axis, and the direction parallel to the bumper is the C-axis.

[0055] Exemplarily, after determining the coordinates of the point cloud of the static target in the global coordinate system, the HCU obtains the NOA (Navigation on Autopilot, autonomous driving technology in urban environments) detection result of the road where the vehicle is located. Among them, the NOA detection result of the road where the vehicle is located indicates whether the road where the vehicle is located is covered by the high-precision map required for NOA; in response to obtaining the NOA detection result of the road where the vehicle is located that the road where the vehicle is located is covered by the high-precision map required for NOA, a static scene is generated according to the high-precision map.

[0056] In a possible implementation, the HCU compares the coordinates of the point cloud of the static target in the global coordinate system with the coordinate area of the area covered by the high-precision map required for NOA in the global coordinate system to determine whether the road where the vehicle is located is covered by the high-precision map required for NOA. If the NOA detection result of the road where the vehicle is located that the road where the vehicle is located is covered by the high-precision map required for NOA is obtained, the HCU generates a static scene according to the high-precision map for subsequent combination with the point cloud of the spliced dynamic target after converting the point cloud in the static scene to the coordinate system of the radar point cloud sensor.

[0057] In step 206, the HCU splices the point cloud of the static target in the global coordinate system to obtain the spliced point cloud of the static target.

[0058] Exemplarily, if the NOA detection result of the road where the vehicle is located that the road where the vehicle is located is not covered by the high-precision map required for NOA is obtained, the HCU splices the point cloud of the static target in the global coordinate system to obtain the spliced point cloud of the static target, including: the HCU splices the point cloud of the static target through a splicing algorithm to obtain the spliced point cloud of the static target. Then, according to the coordinates in the coordinate system of the radar point cloud sensor, the spliced point cloud of the static target is converted back from the global coordinate system to the coordinate system of the radar point cloud sensor.

[0059] In step 207, the HCU combines the spliced point cloud of the dynamic target with the spliced point cloud of the static target to generate the occupancy data of the road where the vehicle is located.

[0060] In a possible implementation, after obtaining the spliced point cloud of the dynamic target and the spliced point cloud of the static target, the HCU combines the spliced point cloud of the dynamic target with the spliced point cloud of the static target to generate the occupancy data of the road where the vehicle is located, including: generating a corresponding static scene according to the spliced point cloud of the static target; restoring the spliced point cloud of the dynamic target to the static scene according to the TrackID and the bounding box information of the dynamic target to obtain the point cloud distribution of the road where the vehicle is located; performing grid division on the road where the vehicle is located; obtaining the occupancy situation of each grid as the occupancy data of the road where the vehicle is located.

[0061] Optionally, through point cloud processing technology, a corresponding static scene is generated based on the point cloud of the spliced static target, and then the point cloud of the spliced dynamic target is restored to the static scene according to the TrackID and the bounding box information of the dynamic target, so as to obtain the point cloud distribution of the road where the vehicle is located. Among them, the point cloud processing technology can be the normal estimation method or the surface reconstruction method.

[0062] Exemplarily, the HCU pre-divides the area of the road where the vehicle is located into multiple grids, obtains the occupancy of each grid, and then determines the occupancy data of the road where the vehicle is located according to the occupancy of each grid. Among them, the size of each grid can be set according to actual needs.

[0063] In a possible implementation manner, the method for obtaining the occupancy of each grid includes, but is not limited to: counting the number of point clouds included in each grid. If the number of point clouds included in any grid is greater than the number threshold, the HCU determines that the grid is occupied; if the number of point clouds included in any grid is less than or equal to the number threshold, the HCU determines that the grid is not occupied. Optionally, the number threshold can be set according to experience.

[0064] Optionally, after obtaining the occupancy data of the road where the vehicle is located, the HCU controls the driving speed and driving direction of the vehicle based on the occupancy data, reduces the risk of the vehicle colliding with obstacles, and thus improves the safety of vehicle driving.

[0065] Exemplarily, the HCU can also capture a video image of the road where the vehicle is located through a camera, and then map the distribution of the point cloud, the bounding box, and the occupancy data to the video image. The video image of the road where the vehicle is located with grids is displayed in real time through an in-vehicle display screen, and the occupied or unoccupied grids are distinguished by different colors, so as to give an intuitive prompt to the driver. Optionally, the camera is installed around the vehicle.

[0066] In the embodiment of the present application, by obtaining the information of the point cloud of consecutive frames received by the radar point cloud sensor, identifying the bounding box information of the dynamic target and the TrackID of the dynamic target, respectively adjusting the point cloud included in the bounding box of each dynamic target, and splicing the adjusted point cloud of the dynamic target, the spliced point cloud of the dynamic target is obtained; by obtaining the coordinates of the point cloud of the static target in the global coordinate system and splicing the point cloud of the static target, the spliced point cloud of the static target is obtained; by combining the spliced point cloud of the dynamic target with the spliced point cloud of the static target, the occupancy data of the road where the vehicle is located is generated, realizing the automatic annotation of the dynamic target and the static target, improving the accuracy of the generated occupancy data of the road where the vehicle is located while controlling the cost, thereby reducing the risk of collision during vehicle driving and ensuring the safety of vehicle driving.

[0067] See Figure 3 , an embodiment of the present application provides an acquisition device for occupancy data, and the device includes:

[0068] The first acquisition module 301 is configured to acquire information of point clouds of consecutive frames received by a radar point cloud sensor, and the point clouds of consecutive frames are composed of points where laser pulses emitted by the radar point cloud sensor are reflected;

[0069] The recognition module 302 is configured to recognize the bounding box of a dynamic target and the TrackID of the dynamic target included in the information of the point clouds of consecutive frames;

[0070] The adjustment module 303 is configured to respectively adjust the point clouds included in the bounding box of each dynamic target to obtain the adjusted point clouds of the dynamic target;

[0071] The first splicing module 304 is configured to splice the adjusted point clouds of the dynamic target to obtain the spliced point clouds of the dynamic target;

[0072] The second acquisition module 305 is configured to acquire the coordinates of the point cloud of a static target in the global coordinate system, and the point cloud of the static target is the remaining point cloud in the point clouds of consecutive frames except for the point clouds included in the bounding box of the dynamic target;

[0073] The second splicing module 306 is configured to splice the point cloud of the static target in the global coordinate system to obtain the spliced point cloud of the static target;

[0074] The combination module 307 is configured to combine the spliced point clouds of the dynamic target with the spliced point clouds of the static target to generate occupancy data of the road where the vehicle is located.

[0075] In a possible implementation manner, the recognition module 302 is configured to detect the information of the point clouds of consecutive frames, recognize the dynamic targets included in each frame and the initial bounding boxes of the dynamic targets; track the movement trajectories of the same dynamic target in each frame, establish the TrackID of each dynamic target; correspond the TrackID to the dynamic targets included in each frame; and correct the initial bounding boxes of the dynamic targets included in each frame based on a preset timing model to obtain the final bounding boxes.

[0076] In a possible implementation manner, the adjustment module 303 is configured to adjust the positions of the point clouds included in the bounding box of each dynamic target so that the center point of the dynamic target is translated to the coordinate origin, and rotate the orientation of the point cloud to a preset standard direction, and the coordinate origin is the origin of the coordinate system of the radar point cloud sensor.

[0077] In a possible implementation, the second acquisition module 305 is configured to acquire the coordinates of the point cloud of the static target in the coordinate system of the radar point cloud sensor; acquire the position and orientation of the radar point cloud sensor in the global coordinate system; and convert the coordinates of the point cloud of the static target in the coordinate system of the radar point cloud sensor into the coordinates of the point cloud of the static target in the global coordinate system based on the position and orientation.

[0078] In a possible implementation, the combination module 307 is configured to generate a corresponding static scene according to the stitched point cloud of the static target; restore the stitched point cloud of the dynamic target to the static scene according to the TrackID and the bounding box information of the dynamic target to obtain the point cloud distribution of the road where the vehicle is located; perform grid division on the road where the vehicle is located; and acquire the occupancy of each grid as the occupancy data of the road where the vehicle is located.

[0079] In a possible implementation, the device further includes: a third acquisition module, configured to acquire the NOA detection result of the road where the vehicle is located, where the NOA detection result of the road where the vehicle is located indicates whether the road where the vehicle is located is covered by the high-precision map required for the NOA of the autonomous driving technology in the urban environment; and a generation module, configured to generate a static scene according to the high-precision map in response to acquiring the NOA detection result of the road where the vehicle is located that is covered by the high-precision map required for NOA.

[0080] In a possible implementation, the combination module 307 is further configured to control the driving speed and driving direction of the vehicle based on the occupancy data.

[0081] This device acquires the information of the point cloud of consecutive frames received by the radar point cloud sensor, identifies the bounding box information and TrackID of the dynamic target, adjusts the point cloud included in the bounding box of each dynamic target respectively, and stitches the adjusted point cloud of the dynamic target to obtain the stitched point cloud of the dynamic target; acquires the coordinates of the point cloud of the static target in the global coordinate system and stitches the point cloud of the static target to obtain the stitched point cloud of the static target; combines the stitched point cloud of the dynamic target with the stitched point cloud of the static target to generate the occupancy data of the road where the vehicle is located, realizes the automatic annotation of the dynamic target and the static target, improves the accuracy of the generated occupancy data of the road where the vehicle is located while controlling the cost, thereby reducing the risk of collision during vehicle driving and ensuring the safety of vehicle driving.

[0082] It should be noted that when the device provided in the above embodiments realizes its functions, only the division of the above functional modules is used for illustration. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device provided in the above embodiments and the method embodiments belong to the same concept. For the specific implementation process, please refer to the method embodiments and will not be elaborated here.

[0083] In an exemplary embodiment, a computer-readable storage medium is further provided. At least one computer program is stored in the computer-readable storage medium. The at least one computer program is loaded and executed by a processor of a computer device so that the computer implements any one of the above methods for obtaining occupancy data.

[0084] In a possible implementation manner, the above computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, etc.

[0085] In an exemplary embodiment, a computer program product or a computer program is further provided. The computer program product or the computer program includes computer instructions. The computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions so that the computer device executes any one of the above methods for obtaining occupancy data.

[0086] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.) and signals involved in this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions. For example, the information of the point cloud, the bounding box of the dynamic target, the TrackID of the dynamic target, and the occupancy data of the road where the vehicle is located involved in this application are all obtained under full authorization.

[0087] It should be understood that "a plurality of" mentioned herein means two or more. "And / or" describes the association relationship of associated objects and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after.

[0088] It should be noted that the terms "first", "second", etc. (if any) in the description and claims of this application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of this application described here can be implemented in an order different from those illustrated or described here. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. On the contrary, they are merely examples of devices and methods consistent with some aspects of this application as detailed in the appended claims.

[0089] The above are only exemplary embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the principles of this application shall be included within the protection scope of this application.

Claims

1. A method for obtaining occupancy data, characterized in that, The method includes: Obtaining information of point clouds of consecutive frames received by a radar point cloud sensor, where the point clouds of the consecutive frames are composed of points reflected by laser pulses emitted by the radar point cloud sensor; Identifying the bounding boxes of dynamic objects and the unique identifier TrackID of the dynamic objects included in the information of the point clouds of the consecutive frames; Adjusting the point clouds included in the bounding box of each dynamic object respectively to obtain the adjusted point clouds of the dynamic objects; Stitching the adjusted point clouds of the dynamic objects to obtain the stitched point clouds of the dynamic objects; Obtaining the coordinates of the point cloud of a static object in the global coordinate system, where the point cloud of the static object is the remaining point cloud in the point clouds of the consecutive frames except for the point clouds included in the bounding box of the dynamic object; Stitching the point cloud of the static object in the global coordinate system to obtain the stitched point cloud of the static object; Combining the stitched point clouds of the dynamic objects with the stitched point clouds of the static objects to generate occupancy data of the road where the vehicle is located.

2. The method according to claim 1, characterized in that, The identifying the bounding boxes of dynamic objects and the TrackID of the dynamic objects included in the information of the point clouds of the consecutive frames includes: Detecting the information of the point clouds of the consecutive frames to identify the dynamic objects included in each frame and the initial bounding boxes of the dynamic objects; Tracking the movement trajectories of the same dynamic object in each frame to establish the TrackID of each dynamic object; Corresponding the TrackID with the dynamic objects included in each frame; Correcting the initial bounding boxes of the dynamic objects included in each frame based on a preset timing model to obtain the final bounding boxes.

3. The method according to claim 1, wherein The adjusting the point clouds included in the bounding box of each dynamic object respectively includes: Adjusting the positions of the point clouds included in the bounding box of each dynamic object so that the center point of the dynamic object is translated to the coordinate origin, and rotating the orientation of the point clouds to a preset standard direction, where the coordinate origin is the origin of the coordinate system of the radar point cloud sensor.

4. The method according to claim 3, wherein The obtaining the coordinates of the point cloud of a static object in the global coordinate system includes: Obtaining the coordinates of the point cloud of the static object in the coordinate system of the radar point cloud sensor; Obtaining the position and orientation of the radar point cloud sensor in the global coordinate system; Converting the coordinates of the point cloud of the static object in the coordinate system of the radar point cloud sensor to the coordinates of the point cloud of the static object in the global coordinate system based on the position and the orientation.

5. The method according to claim 1, characterized in that, The combining the stitched point clouds of the dynamic objects with the stitched point clouds of the static objects to generate occupancy data of the road where the vehicle is located includes: Generating a corresponding static scene according to the stitched point cloud of the static object; Restoring the stitched point clouds of the dynamic objects to the static scene according to the TrackID and the bounding box information of the dynamic objects to obtain the point cloud distribution of the road where the vehicle is located; Performing grid division on the road where the vehicle is located; Obtaining the occupancy situation of each grid as the occupancy data of the road where the vehicle is located.

6. The method according to claim 5, wherein The method further includes: Obtain the NOA detection result of the road where the vehicle is located, and the NOA detection result of the road where the vehicle is located indicates whether the road where the vehicle is located is covered by the high-precision map required for the NOA of autonomous driving technology in the urban environment; In response to obtaining the NOA detection result of the road where the vehicle is located that the road where the vehicle is located is covered by the high-precision map required for the NOA, generate the static scene according to the high-precision map.

7. The method according to claim 1, wherein After generating the occupancy data of the road where the vehicle is located, it further includes: Control the driving speed and driving direction of the vehicle based on the occupancy data.

8. An acquisition device for occupied data, characterized in that, The device includes: A first acquisition module for acquiring information of point clouds of consecutive frames received by a radar point cloud sensor, and the point clouds of the consecutive frames are composed of points where laser pulses emitted by the radar point cloud sensor are reflected; An identification module for identifying the bounding box of the dynamic target and the TrackID of the dynamic target included in the information of the point clouds of the consecutive frames; An adjustment module for respectively adjusting the point clouds included in the bounding box of each dynamic target to obtain the adjusted point clouds of the dynamic target; A first splicing module for splicing the adjusted point clouds of the dynamic target to obtain the spliced point clouds of the dynamic target; A second acquisition module for acquiring the coordinates of the point cloud of the static target in the global coordinate system, and the point cloud of the static target is the remaining point cloud in the point clouds of the consecutive frames except the point clouds included in the bounding box of the dynamic target; A second splicing module for splicing the point cloud of the static target in the global coordinate system to obtain the spliced point cloud of the static target; A combination module for combining the spliced point clouds of the dynamic target with the spliced point clouds of the static target to generate the occupancy data of the road where the vehicle is located.

9. A computer program product, the computer program product includes computer instructions, and when the computer instructions are executed by a processor, the steps of the method for obtaining occupancy data as described in any one of claims 1 to 7 are implemented.

10. A non - transitory computer - readable storage medium, characterized in that, A computer program is stored in the computer-readable storage medium, and the computer program is loaded and executed by a processor to implement the method for obtaining occupancy data as described in any one of claims 1 to 7.

Images