Static obstacle detection method, device, equipment and computer storage medium
By adding a feature vector similarity judgment based on consecutive frames to the obstacle detection model, the problem that existing models have difficulty detecting obstacles outside the training data category is solved, and more accurate obstacle detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-13
- Publication Date
- 2026-03-31
AI Technical Summary
Existing obstacle detection models struggle to detect obstacles outside the categories covered by the training data, resulting in incomplete detection results.
By acquiring point cloud data and mapping it to a grid map, after determining that there are obstacles in the grid, the similarity of feature vectors between consecutive frames is used to determine whether the obstacle is a static obstacle. A judgment method based on consecutive frames is added to improve detection performance.
It significantly improves the detection performance of obstacles outside the training data category and enhances the accuracy of obstacle detection.
Smart Images
Figure CN115223140B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of autonomous driving technology, and in particular to a static obstacle detection method, device, equipment, and computer storage medium. Background Technology
[0002] With the development of computer and artificial intelligence technologies, autonomous driving technology has rapidly advanced in order to change the way humans travel and transport goods. In autonomous driving technology, the system uses sensors to perceive the surrounding environment of the vehicle and identifies obstacles based on the perception results to achieve functions such as path planning.
[0003] Common obstacle detection methods mainly use deep learning technology to divide samples into training and testing sets. The obstacle detection model is trained based on the training set to obtain a trained obstacle detection model, and then the trained obstacle detection model is used to detect obstacles.
[0004] However, current obstacle detection models trained on training sets are unable to detect obstacles outside the categories of the training data, resulting in incomplete obstacle detection results. Summary of the Invention
[0005] This application provides a static obstacle detection method, apparatus, device, and computer storage medium to improve the detection performance of obstacles outside the training data category.
[0006] On the one hand, this application provides a static obstacle detection method, including:
[0007] Obtain point cloud data at the current moment and map the point cloud data onto a first grid map. Obtain grid information for each first grid cell in the first grid map. The first grid map is a map in the vehicle coordinate system. The grid information includes the height information of the points and the number of points.
[0008] After determining that there is an obstacle in the first target grid based on the grid information of each first grid, the feature vector of the first target grid is determined.
[0009] The second target grid corresponding to the first target grid is obtained from the second grid map, and the feature vector of the first target grid is associated with and stored with the current time and the position information of the second target grid; the second grid map is a map in the real world coordinate system;
[0010] The feature vectors of the second target grid at different times are obtained, and the obstacle is determined to be a static obstacle based on the feature vectors of the second target grid at different times.
[0011] Optionally, determining the feature vector of the first target raster includes:
[0012] The feature vector of the first target grid is determined based on the grid information of the first target grid and the grid information of the first grid adjacent to the first target grid.
[0013] Optionally, determining the feature vector of the first target grid based on the grid information of the first target grid and the grid information of the first grid adjacent to the first target grid includes:
[0014] Based on the grid information of the first target grid, the maximum height of the points in the first target grid and the number of points in the first target grid are obtained, and the maximum height and the number of points corresponding to the first target grid are used as the first sub-feature vector.
[0015] Based on the grid information of each first grid cell, the maximum height of the point in each first grid cell and the number of points in each first grid cell are obtained, and the maximum height and the number of points corresponding to each first grid cell are used as the second sub-feature vector;
[0016] The feature vector of the first target raster is determined based on the first sub-feature vector and multiple second sub-feature vectors.
[0017] Optionally, determining whether the obstacle is a static obstacle based on the feature vectors of the second target grid at different times includes:
[0018] Based on the similarity between the feature vector of the second target grid at the current moment and the feature vector of the second target grid at a historical moment, it is determined whether the obstacle is a static obstacle.
[0019] Optionally, determining whether the obstacle is a static obstacle based on the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at a historical time includes:
[0020] Based on the similarity between the feature vector of the second target grid at the current moment and the feature vector of the second target grid at each historical moment, determine the number of similarities greater than a preset threshold;
[0021] If the number is greater than the preset number, then the obstacle is determined to be a static obstacle.
[0022] Optionally, obtaining the feature vectors of the second target raster at different times includes:
[0023] Using the location information of the second target grid and the preset time period as indexes, the feature vectors of the second target grid at different times are obtained from the feature vectors associated with and stored by the second target grid.
[0024] Optionally, determining the presence of an obstacle in the first target grid based on the grid information of each first grid cell includes:
[0025] Based on the height information of each point in the first grid, determine the number of points whose height value in the height information of the first grid is greater than a preset threshold;
[0026] If the quantity is greater than a preset value, then the first grid is determined to be the first target grid, and there is an obstacle in the first target grid.
[0027] Secondly, this application provides a static obstacle detection device, comprising:
[0028] The first acquisition module is used to acquire the point cloud data at the current moment;
[0029] The first processing module is used to map the point cloud data to a first grid map to obtain grid information of each first grid in the first grid map. The first grid map is a map in a vehicle coordinate system. The grid information includes the height information of the points mapped to the grid and the number of points.
[0030] The first determining module is used to determine the feature vector of the first target grid after determining that there is an obstacle in the first target grid based on the grid information of each first grid.
[0031] The second acquisition module is used to acquire the second target grid corresponding to the first target grid in the second grid map, wherein the second grid map is a map in the real world coordinate system;
[0032] The first storage module is used to associate and store the feature vector of the first target grid with the current time and the position information of the second target grid.
[0033] The third acquisition module is used to acquire the feature vectors of the second target raster at different times;
[0034] The second determining module is used to determine whether the obstacle is a static obstacle based on the feature vectors of the second target grid at different times.
[0035] Optionally, the first determining module is specifically used to determine the feature vector of the first target grid based on the grid information of the first target grid and the grid information of the first grid adjacent to the first target grid.
[0036] Optionally, the first determining module is further configured to: obtain the maximum height of the points in the first target grid and the number of points in the first target grid based on the grid information of the first target grid; use the maximum height and the number of points corresponding to the first target grid as a first sub-feature vector; obtain the maximum height of the points in each first grid and the number of points in each first grid based on the grid information of each first grid; use the maximum height and the number of points corresponding to each first grid as a second sub-feature vector; and determine the feature vector of the first target grid based on the first sub-feature vector and multiple second sub-feature vectors.
[0037] Optionally, the second determining module is specifically used to determine whether the obstacle is a static obstacle based on the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at a historical time.
[0038] Optionally, the second determining module is further configured to determine the number of similarities greater than a preset threshold based on the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at each historical time; if the number is greater than the preset number, then the obstacle is determined to be a static obstacle.
[0039] Optionally, the third acquisition module is specifically used to acquire the feature vectors of the second target grid at different times from the feature vectors associated with and stored by the second target grid, using the location information of the second target grid and the preset time period as indexes.
[0040] Optionally, the first determining module is further configured to determine, based on the height information of each point in the first grid, the number of points whose height value in the height information of the first grid is greater than a preset threshold; if the number is greater than the preset value, then the first grid is determined to be a first target grid, and there is an obstacle in the first target grid.
[0041] Thirdly, this application provides a static obstacle detection device, comprising:
[0042] Memory;
[0043] processor;
[0044] The memory stores computer-executed instructions;
[0045] The processor executes computer execution instructions stored in the memory to implement the static obstacle detection method as described in the first aspect and various possible implementations of the first aspect above.
[0046] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the static obstacle detection method as described in the first aspect and various possible implementations of the first aspect above.
[0047] The static obstacle detection method provided in this embodiment acquires point cloud data at the current moment and maps the point cloud data to a first grid map. It then acquires the grid information of each first grid cell in the first grid map, which is a map in a vehicle coordinate system. The grid information includes the height information and the number of points. After determining that an obstacle exists in a first target grid cell based on the grid information of each first grid cell, it determines the feature vector of the first target grid cell. Next, it acquires a second target grid cell corresponding to the first target grid cell in a second grid map and associates and stores the feature vector of the first target grid cell with the current moment and the position information of the second target grid cell. The second grid map is a map in a real-world coordinate system. Finally, it acquires the feature vectors of the second target grid cell at different times and determines whether the obstacle is a static obstacle based on the feature vectors of the second target grid cell at different times. Compared with existing deep learning model-based techniques, the static obstacle detection method provided in this embodiment adds a method based on previous and subsequent frames to determine whether an obstacle is a static obstacle, thereby significantly improving the detection performance for obstacles outside the training data categories. Attached Figure Description
[0048] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0049] Figure 1 A schematic diagram of a scenario for the static obstacle detection method provided in this application;
[0050] Figure 2 The flowchart of the static obstacle detection method provided in this application Figure 1 ;
[0051] Figure 3 The flowchart of the static obstacle detection method provided in this application Figure 2 ;
[0052] Figure 4 The flowchart for obstacle detection and feature extraction provided in this application;
[0053] Figure 5 Please provide the feature vectors and raster map diagrams for this application.
[0054] Figure 6 The static obstacle determination flowchart provided in this application;
[0055] Figure 7 A schematic diagram of the static obstacle detection device provided in this application;
[0056] Figure 8 This is a structural schematic diagram of the static obstacle detection device provided in this application.
[0057] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation
[0058] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0059] Autonomous driving refers to the technology that enables a car to drive normally without human control. This technology is becoming increasingly mature, and because autonomous vehicles can save more human resources, this technology is receiving more and more attention.
[0060] In autonomous driving technology, obstacle detection is particularly important for self-driving vehicles, and currently, obstacle detection models are mainly used to identify obstacles. During the training of an obstacle detection model, the first step is to acquire sample data, which can be obtained from relevant data within the obstacle scene. Figure 1 This is a schematic diagram illustrating a scenario for the static obstacle detection method provided in an embodiment of this application. Figure 1 As shown, there are obstacles such as vehicles, streetlights, and boxes around the autonomous vehicle. During the autonomous vehicle's operation, it acquires point cloud data and uses existing deep learning models to detect obstacles in the point cloud data in order to identify them.
[0061] However, in real-world scenarios, there are often other obstacles beyond the training dataset, such as railings, walls, streetlights, green belts, and irregular stone blocks. Existing deep learning models struggle to detect these obstacles outside the training data categories, resulting in incomplete detection results. Therefore, how to detect obstacles outside the training dataset is a problem that needs to be solved.
[0062] The method provided in this application, after determining that an obstacle exists in a target grid, determines whether the obstacle is a static obstacle based on the similarity of the target grid across several consecutive frames. Compared with existing deep learning model-based techniques, the method provided in this application adds a method based on consecutive frames to determine whether an obstacle is a static obstacle, thereby significantly improving the detection performance for obstacles outside the training data categories.
[0063] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.
[0064] Figure 2 The flowchart of the static obstacle detection method shown in the embodiments of this application Figure 1 .like Figure 2 As shown, the static obstacle detection method provided in this embodiment includes:
[0065] S101: Obtain the point cloud data at the current moment, and map the point cloud data to the first grid map. Obtain the grid information of each first grid in the first grid map. The first grid map is a map in the vehicle coordinate system. The grid information includes the height information of the points and the number of points.
[0066] Point cloud data refers to a set of vectors in a three-dimensional coordinate system. The first grid map is a grid map with vehicles as reference points, which is neatly divided into several grids with fixed lengths and widths.
[0067] The height of a point refers to its elevation above the ground. In this step, it is determined that existing technologies can be used to determine the height information of the point. For example, the RaySegment method can be used to segment the ground information, and then the height information of each point can be obtained based on the ground information and the point cloud information. This application does not impose any particular limitations on the implementation method for obtaining the height information of points.
[0068] In this step, for example, the position of each point in the first grid map can be determined based on the horizontal and vertical coordinates of the point cloud data, thereby mapping the point cloud data to the first grid map. After mapping the point cloud data to the first grid map, the total number of points included in each first grid and the height information of the points in each first grid can be obtained. This application embodiment does not impose any particular limitation on the implementation method of obtaining the grid information of each first grid.
[0069] S102: After determining that there is an obstacle in the first target grid based on the grid information of each first grid, the feature vector of the first target grid is determined.
[0070] The first target grid is the grid containing obstacles. After obtaining the grid information of each first grid, the first target grid containing obstacles can be determined based on the grid information of each first grid. This application embodiment does not impose any particular limitation on the implementation method of determining the presence of obstacles in the first target grid.
[0071] The feature vector is used to characterize the current features of the first target grid for subsequent matching and comparison. After identifying the first target grid containing an obstacle, the feature vector of that first target grid can be determined. This application does not impose any particular limitations on the implementation method for determining the feature vector of the first target grid.
[0072] S103: Obtain the second target grid corresponding to the first target grid in the second grid map, and associate and store the feature vector of the first target grid with the current time and the position information of the second target grid; the second grid map is a map in the real world coordinate system.
[0073] Since the vehicle is continuously moving, the position of the first target grid relative to the vehicle changes at different times. Therefore, it is necessary to obtain the second target grid corresponding to the first target grid from the second grid map. The second grid map is a map in a real-world coordinate system, such as a map in a UTM coordinate system. The position information of the second target grid is the same at different times. This application embodiment does not impose any particular restrictions on the implementation method of obtaining the second target grid.
[0074] For example, the location range of the first target grid on the second grid map can be determined based on the horizontal and vertical coordinates of the first target grid, and then the second target grid corresponding to the first target grid can be determined.
[0075] For example, the feature vector of the first target grid can be stored using the current time and the position information of the second target grid as an index. That is, the feature vector is associated with the corresponding time and the position information in the real-world coordinate system corresponding to the feature vector and stored accordingly.
[0076] S104: Obtain the feature vectors of the second target grid at different times, and determine whether the obstacle is a static obstacle based on the feature vectors of the second target grid at different times.
[0077] The process involves acquiring feature vectors of the second target grid at different times, and determining whether an obstacle is a static obstacle based on these feature vectors. This application does not impose any particular limitations on the implementation method of determining whether an obstacle is a static obstacle based on the feature vectors of the second target grid at different times.
[0078] For example, the similarity of feature vectors at different times can be used to determine whether an obstacle is a static obstacle.
[0079] The static obstacle detection method provided in this embodiment includes: acquiring point cloud data at the current moment and mapping the point cloud data to a first grid map; acquiring grid information of each first grid cell in the first grid map, wherein the first grid map is a map in a vehicle coordinate system, and the grid information includes the height information and the number of points; after determining that an obstacle exists in a first target grid cell based on the grid information of each first grid cell, determining the feature vector of the first target grid cell; acquiring a second target grid cell corresponding to the first target grid cell in a second grid map, and storing the feature vector of the first target grid cell in association with the current moment and the position information of the second target grid cell; wherein the second grid map is a map in a real-world coordinate system; acquiring the feature vector of the second target grid cell at different moments, and determining whether the obstacle is a static obstacle based on the feature vector of the second target grid cell at different moments. Compared with existing deep learning model-based techniques, the static obstacle detection method provided in this embodiment adds a method based on previous and subsequent frames to determine whether an obstacle is a static obstacle in addition to the deep learning model, thereby significantly improving the detection performance of obstacles outside the training data category.
[0080] Figure 3 The flowchart of the static obstacle detection method shown in the embodiments of this application Figure 2 , Figure 4 This is a flowchart of obstacle detection and feature extraction. Figure 5 This is a schematic diagram of feature vectors and a raster map. Figure 6 This is a flowchart for static obstacle detection. (Combined with...) Figures 3 to 6 This embodiment, based on the above embodiments, provides a detailed explanation of how to determine whether an obstacle is a static obstacle.
[0081] S201: Obtain the point cloud data at the current moment, and map the point cloud data to the first grid map. Obtain the grid information of each first grid in the first grid map. The first grid map is a map in the vehicle coordinate system. The grid information includes the height information of the points and the number of points.
[0082] Step S201 is similar to step S101 above, and will not be repeated here.
[0083] S202: Based on the height information of each point in the first grid, determine the number of points in the height information of the first grid whose height value is greater than a preset threshold.
[0084] S203: If the quantity is greater than a preset value, then the first grid is determined to be the first target grid, and there is an obstacle in the first target grid.
[0085] Combination Figure 4 As can be seen, after mapping the point cloud data onto the first grid map, obstacle detection is performed. In this step, the height value of each point in the first grid is compared with a preset threshold to determine the number of points whose height value is greater than the preset threshold. When the number of points is greater than the preset value, the first grid can be determined as a target grid containing obstacles.
[0086] Those skilled in the art will understand that this application determines the presence of obstacles based on the height information of each point. Compared with the prior art, which uses a trained obstacle detection model to detect obstacles, this application can detect not only obstacles of the conventional category, but also obstacles outside the categories of the training data, thereby improving the accuracy of obstacle detection.
[0087] S204: Determine the feature vector of the first target grid based on the grid information of the first target grid and the grid information of the first grid adjacent to the first target grid.
[0088] In this step, a feature vector needs to be set for the first target raster for subsequent matching and comparison. This application does not impose any particular limitations on the implementation method for determining the feature vector of the first target raster.
[0089] For example, one possible implementation is given here. Based on the grid information of the first target grid, the maximum height of the points in the first target grid and the number of points in the first target grid are obtained, and the maximum height and the number of points corresponding to the first target grid are used as a first sub-feature vector; based on the grid information of each first grid, the maximum height of the points in each first grid and the number of points in each first grid are obtained, and the maximum height and the number of points corresponding to each first grid are used as a second sub-feature vector; based on the first sub-feature vector and multiple second sub-feature vectors, the feature vector of the first target grid is determined.
[0090] For example, in combination Figure 5 It can be seen that the first target grid is G. i,j The first grid cells adjacent to the first target grid cells are G. i-1,j-1 G i-1,j ...G i+1,j+1The second grid adjacent to the first target grid can be any grid centered on the first target grid. The feature vector can be 18 bits. The first two bits of the feature vector are used to store the grid information of the first target grid: bit 0 can be the maximum height of the points in the first target grid; bit 1 can be the total number of points in the first target grid. Bits 2-17 of the feature vector are used to store the grid information of the first grid adjacent to the first target grid, where every two bits represent the grid information of one first grid, which can be the maximum height of the points in the corresponding first grid and the total number of points, respectively.
[0091] Those skilled in the art will understand that determining the feature vector of the first target grid by using the grid information of the first target grid and the grid information of the adjacent first grids can more accurately express the features of the first target grid, thereby making the subsequent comparison results more accurate.
[0092] S205: Obtain the second target grid corresponding to the first target grid in the second grid map, and associate and store the feature vector of the first target grid with the current time and the position information of the second target grid; the second grid map is a map in the real world coordinate system.
[0093] Step S205 is similar to step S103 above, and will not be described again here.
[0094] S206: Using the location information of the second target grid and the preset time period as indexes, obtain the feature vectors of the second target grid at different times from the feature vectors associated with the second target grid.
[0095] Since the location information, storage time point, and feature vector of the second target grid are related, the feature vector of the second target grid at different times within a preset time period can be obtained using the location information and storage time point of the second target grid as indexes. This embodiment does not impose special restrictions on the preset time period; different preset time periods can be set according to actual circumstances.
[0096] S207: Determine whether the obstacle is a static obstacle based on the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at a historical time.
[0097] The feature vector of the second target grid at a historical moment can include feature vectors of second target grids at multiple different moments. After obtaining the feature vectors of the second target grid at different moments, the similarity between the feature vector of the second target grid at the current moment and the feature vectors of the second target grid at historical moments can be calculated to determine whether the obstacle in the first target grid is a static obstacle. This application does not impose any particular limitation on the implementation method for determining whether an obstacle is a static obstacle.
[0098] For example, one possible implementation is given here. Based on the similarity between the feature vector of the second target grid at the current moment and the feature vector of the second target grid at each historical moment, the number of similarities greater than a preset threshold is determined; if the number is greater than the preset number, the obstacle is determined to be a static obstacle.
[0099] Combination Figure 6 It can be seen that the feature vector of the second target grid at the current time can be represented by T. i This means that the feature vector of the second target raster at the searched historical moment can be T. i-1 T i-2 ...T i-k After obtaining the feature vectors of the second target raster at different times, the feature vector T of the second target raster at the current time is calculated. i The feature vector T of the second target raster at a historical moment i-1 T i-2 ...T i-k The similarity is counted, and the number of similarities greater than a preset threshold is counted. If the number of similarities greater than the preset threshold is greater than or equal to a preset number N, then the obstacle in the first target grid can be determined to be a static obstacle; if the number of similarities greater than the preset threshold is less than the preset number N, then the obstacle in the first target grid can be determined to be a non-static obstacle.
[0100] The static obstacle detection method provided in this embodiment determines the presence of obstacles based on the height information of each point in the point cloud data. Compared to existing technologies that use a trained obstacle detection model, this method can detect not only obstacles of the common categories but also obstacles outside the training data categories, thereby improving the accuracy of obstacle detection. Furthermore, after detecting an obstacle, it utilizes the characteristic that static obstacles change relatively little between consecutive frames to determine whether the detected obstacle is truly static. Compared to existing deep learning model-based techniques, this method adds a frame-based approach to determine whether an obstacle is static, significantly improving the detection performance for obstacles outside the training data categories.
[0101] Figure 7 This is a structural schematic diagram of the static obstacle detection device provided in this application. Figure 7 As shown, this application provides a static obstacle detection device, the static obstacle detection device 300 including:
[0102] The first acquisition module 301 is used to acquire point cloud data at the current moment;
[0103] The first processing module 302 is used to map the point cloud data to a first grid map to obtain grid information of each first grid in the first grid map. The first grid map is a map in a vehicle coordinate system. The grid information includes the height information of the points mapped to the grid and the number of points.
[0104] The first determining module 303 is used to determine the feature vector of the first target grid after determining that there is an obstacle in the first target grid based on the grid information of each first grid.
[0105] The second acquisition module 304 is used to acquire the second target grid corresponding to the first target grid in the second grid map, wherein the second grid map is a map in the real world coordinate system;
[0106] The first storage module 305 is used to associate and store the feature vector of the first target grid with the current time and the position information of the second target grid.
[0107] The third acquisition module 306 is used to acquire the feature vectors of the second target raster at different times;
[0108] The second determining module 307 is used to determine whether the obstacle is a static obstacle based on the feature vector of the second target grid at different times.
[0109] Optionally, the first determining module 303 is specifically used to determine the feature vector of the first target grid based on the grid information of the first target grid and the grid information of the first grid adjacent to the first target grid.
[0110] Optionally, the first determining module 303 is further configured to: obtain the maximum height of the midpoints of the first target grid and the number of midpoints of the first target grid based on the grid information of the first target grid; use the maximum height and the number of midpoints of the first target grid as a first sub-feature vector; obtain the maximum height of the midpoints of each first grid and the number of midpoints of each first grid based on the grid information of each first grid; use the maximum height and the number of midpoints of each first grid as a second sub-feature vector; and determine the feature vector of the first target grid based on the first sub-feature vector and multiple second sub-feature vectors.
[0111] Optionally, the second determining module 307 is specifically used to determine whether the obstacle is a static obstacle based on the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at a historical time.
[0112] Optionally, the second determining module 307 is further configured to determine the number of similarities greater than a preset threshold based on the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at each historical time; if the number is greater than the preset number, then the obstacle is determined to be a static obstacle.
[0113] Optionally, the third acquisition module 306 is specifically used to acquire the feature vectors of the second target grid at different times from the feature vectors associated with and stored by the second target grid, using the location information of the second target grid and the preset time period as indexes.
[0114] Optionally, the first determining module 303 is further configured to determine, based on the height information of each point in the first grid, the number of points whose height values in the height information of the first grid are greater than a preset threshold; if the number is greater than the preset value, then the first grid is determined to be a first target grid, and there is an obstacle in the first target grid.
[0115] Figure 8 This is a structural schematic diagram of the static obstacle detection device provided in this application. Figure 8 As shown, this application provides a static obstacle detection device 400, which includes a receiver 401, a transmitter 402, a processor 403, and a memory 404.
[0116] Receiver 401 is used to receive instructions and data;
[0117] Transmitter 402 is used to send commands and data;
[0118] Memory 404 is used to store instructions executed by the computer;
[0119] The processor 403 is used to execute computer execution instructions stored in the memory 404 to implement the various steps of the static obstacle detection method in the above embodiments. For details, please refer to the relevant descriptions in the foregoing embodiments of the static obstacle detection method.
[0120] Alternatively, the memory 404 can be either standalone or integrated with the processor 403.
[0121] When the memory 404 is set up independently, the electronic device also includes a bus for connecting the memory 404 and the processor 403.
[0122] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the static obstacle detection method performed by the static obstacle detection device described above.
[0123] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.
[0124] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.
[0125] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.
Claims
1. A static obstacle detection method characterized by, The method comprises the following steps: acquiring point cloud data at a current time and mapping the point cloud data to a first grid map, acquiring grid information of each first grid in the first grid map, the first grid map being a map in a vehicle coordinate system, the grid information comprising height information of points and a number of points; after determining that an obstacle exists in a first target grid according to the grid information of each first grid, determining a feature vector of the first target grid according to the grid information of the first target grid and the grid information of a first grid adjacent to the first target grid; acquiring a second target grid corresponding to the first target grid in a second grid map and storing the feature vector of the first target grid in association with a current time and position information of the second target grid, the second grid map being a map in a real-world coordinate system; acquiring feature vectors of the second target grid at different times and determining whether the obstacle is a static obstacle according to the feature vectors of the second target grid at different times; wherein, the feature vector of the first target grid is determined according to the grid information of the first target grid and the grid information of the first grid adjacent to the first target grid, comprising: the maximum height of points in the first target grid and the number of points in the first target grid are acquired according to the grid information of the first target grid, and the maximum height and the number of points corresponding to the first target grid are taken as a first sub-feature vector; the maximum height of points in each first grid and the number of points in each first grid are acquired according to the grid information of each first grid, and the maximum height and the number of points corresponding to each first grid are taken as a second sub-feature vector; the feature vector of the first target grid is determined according to the first sub-feature vector and a plurality of second sub-feature vectors.
2. The method of claim 1, wherein, the determination of whether the obstacle is a static obstacle according to the feature vectors of the second target grid at different times, comprising: whether the obstacle is a static obstacle is determined according to the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at a historical time.
3. The method of claim 2, wherein, the determination of whether the obstacle is a static obstacle according to the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at a historical time, comprising: the number of similarities greater than a preset threshold is determined according to the similarity between the feature vector of the second target grid at the current time and the feature vector of the second target grid at each historical time; if the number is greater than a preset number, it is determined that the obstacle is a static obstacle.
4. The method of claim 1, wherein, the acquisition of the feature vectors of the second target grid at different times, comprising: the feature vectors of the second target grid at different times are acquired from the feature vectors stored in association with the second target grid, with the position information of the second target grid and a preset time period as indexes.
5. The method of claim 1, wherein, the determination of whether an obstacle exists in a first target grid according to the grid information of each first grid, comprising: the number of points with a height value greater than a preset threshold in the height information in the first grid is determined according to the height information of each point in the first grid. If the number is greater than a preset value, it is determined that the first grid is a first target grid, and the first target grid has an obstacle.
6. A static obstacle detection apparatus characterized by comprising: The application relates to a static obstacle detection method and device. The first acquisition module is configured to acquire point cloud data at a current time. The first processing module is configured to map the point cloud data to a first grid map to obtain grid information of each first grid in the first grid map, the first grid map being a map in a vehicle coordinate system, and the grid information including height information and a number of points mapped into a grid. The first determination module is configured to determine a feature vector of the first target grid according to grid information of the first target grid and grid information of first grids adjacent to the first target grid after determining that the first target grid has an obstacle according to the grid information of each first grid. The second acquisition module is configured to acquire a second target grid corresponding to the first target grid in a second grid map, the second grid map being a map in a real world coordinate system. The first storage module is configured to store the feature vector of the first target grid in association with a current time and position information of the second target grid. The third acquisition module is configured to acquire feature vectors of the second target grid at different times. The second determination module is configured to determine whether the obstacle is a static obstacle according to the feature vectors of the second target grid at different times.
7. A static obstacle detection apparatus characterized by comprising: The application relates to a static obstacle detection method and device. The computer storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the static obstacle detection method. The computer program product contains a computer program, and the computer program is executed by the processor to implement the method in any one of claims 1 to 5. The computer program product contains a computer program, and the computer program is executed by the processor to implement the method in any one of claims 1 to 5. 8. A computer storage medium, characterized in that, 9. A computer program product, characterised in that,
Citation Information
Patent Citations
Static obstacle detection method, device and equipment and storage medium
CN111339996A