Occupancy grid tracking method, device, medium and product

Abstract

The embodiments of the present application relate to the field of intelligent transportation technology, and disclose a method, device, medium and product for tracking an occupied grid. The method for tracking an occupied grid includes: obtaining first feature information based on an observation grid and second feature information based on a prediction grid; wherein the first feature information includes the average position of the point cloud in the observation grid and the number of first static point clouds, and the second feature information includes the predicted position coordinates of the prediction grid and the number of second static point clouds; according to the first feature information and the second feature information, determining an updated target prediction grid; the prediction grid that has no tracking value has been deleted from the target prediction grid; and tracking of the occupied grid is achieved according to the target prediction grid. It can at least be used to solve the technical problems in the related technology that the entire grid map needs to be maintained, a large amount of grid data processing is involved, the calculation complexity is high and time-consuming, and the tracking efficiency of the occupied grid is low.

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation technology, and in particular to an occupation grid tracking method, device, medium and product. Background Art

[0002] Environmental perception technology is crucial in fields such as autonomous driving and intelligent transportation. Traditional environmental perception technology mainly relies on sensors such as LiDAR and millimeter-wave radar. Among them, millimeter-wave radar uses electromagnetic waves in the millimeter-wave frequency band to measure the distance, speed and angle between the target object and the radar by transmitting and receiving echo signals. Compared with LiDAR, millimeter-wave radar has a longer range. For example, the detection distance of millimeter-wave radar can reach up to 200 meters, and it can output occupancy grids in a longer range. Among them, the occupancy grid is a data structure used to represent the occupancy of environmental space. It can divide the environmental space around the vehicle into a series of grid cells. Each grid cell contains a value, which indicates the probability of the grid cell being occupied.

[0003] However, the inventors have found that there are at least the following technical problems in the related art: in the related art, the entire grid map needs to be maintained, involving a large amount of grid data processing, the calculation complexity is high and time-consuming, and the tracking efficiency of the occupied grid is low. Summary of the invention

[0004] One purpose of the present application is to provide a method, device, medium and product for tracking an occupied grid, at least to solve the technical problems in the related art that the entire grid map needs to be maintained, a large amount of grid data processing is involved, the calculation complexity is high and time-consuming, and the tracking efficiency of the occupied grid is low.

[0005] To achieve the above objectives, some embodiments of the present application provide the following aspects:

[0006] In the first aspect, some embodiments of the present application also provide a method for tracking an occupied grid, the method comprising: obtaining first feature information based on an observation grid and second feature information based on a prediction grid; wherein the first feature information comprises the average position of a point cloud in the observation grid and the number of first static point clouds, and the second feature information comprises the predicted position coordinates of the prediction grid and the number of second static point clouds; determining an updated target prediction grid based on the first feature information and the second feature information; deleting prediction grids that have no tracking value from the target prediction grid; and tracking the occupied grid based on the target prediction grid.

[0007] In a second aspect, some embodiments of the present application further provide an electronic device, comprising: one or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, cause the processor to perform the steps of the method described above.

[0008] In a third aspect, some embodiments of the present application further provide a computer-readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method as described above.

[0009] In a fourth aspect, some embodiments of the present application further provide a computer program product, comprising a computer program / instruction, which implements the steps of the method described above when executed by a processor.

[0010] Compared with the related art, in the solution provided by the embodiment of the present application, by obtaining the first feature information based on the observation grid and the second feature information based on the prediction grid; wherein the first feature information includes the average position of the point cloud in the observation grid and the number of the first static point cloud, and the second feature information includes the predicted position coordinates of the prediction grid and the number of the second static point cloud, then according to the first feature information and the second feature information, the updated target prediction grid is determined; the prediction grid that has no tracking value has been deleted from the target prediction grid, and the tracking of the occupied grid is realized according to the target prediction grid. Since the prediction grid that has no tracking value has been deleted from the target prediction grid, it is not necessary to maintain the entire grid map, which reduces the high and time-consuming calculation complexity and improves the tracking efficiency of the occupied grid. By tracking the occupied grid based on the target prediction grid, the entire system can realize continuous monitoring of the position of static objects. It can be applied to timely discover obstacles on the road and reasonably plan driving routes in autonomous driving scenarios, which helps to improve the accuracy of the system's environmental perception and the rationality of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0012] Figure 1 An exemplary flow chart of an occupation grid tracking method provided according to some embodiments of the present application;

[0013] Figure 2 This is an exemplary effect diagram of an occupation grid tracking method provided according to some embodiments of the present application;

[0014] Figure 3 This is a schematic diagram of an exemplary structure of an electronic device provided according to some embodiments of the present application. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0016] First embodiment

[0017] The first embodiment of the present application relates to a method for tracking an occupied grid. Figure 1 As shown, the method may include the following steps:

[0018] Step S101, obtaining first feature information based on an observation grid and second feature information based on a prediction grid; wherein the first feature information includes an average position of a point cloud in the observation grid and a first number of static point clouds, and the second feature information includes a predicted position coordinate of the prediction grid and a second number of static point clouds;

[0019] Step S102, determining an updated target prediction grid according to the first feature information and the second feature information; the prediction grid having no tracking value has been deleted from the target prediction grid;

[0020] Step S103: tracking the occupied grid according to the target predicted grid.

[0021] The method can be applied to a system, and the system can be installed in an electronic device. The following describes each of the above steps in detail.

[0022] For step S101, specifically, in some examples, the observation grid is used to represent the units after the spatial range that the laser radar device can observe is divided. The laser radar device can be a 4D millimeter wave radar. Exemplarily, after the point cloud data acquired by the 4D millimeter wave radar is projected onto the grid map to form an observation grid, the average position coordinates of these point clouds in three-dimensional space can be calculated for each point cloud in the observation grid, which is used to represent the central trend position of the point cloud in the observation grid and reflect the possible position center of the static object in the observation grid. The first number of static point clouds refers to the statistical value of the number of point clouds determined to belong to static objects in the observation grid.

[0023] Specifically, in some examples, the predicted grid is an estimate of the grid position at the next moment based on the known grid position of the previous frame and the vehicle's own motion information (such as the vehicle's motion speed, yaw rate, coordinate position change value between frames, etc.). The predicted position coordinates are the key results of the prediction of the dynamic change of the grid position in the entire grid tracking algorithm, and will be subsequently correlated, compared, and fused and updated with the position information of the observed grid. The second static point cloud number refers to the statistical value of the number of point clouds that are determined to belong to static objects in the predicted grid.

[0024] For step S102, specifically, in some examples, the target prediction grid can be determined by using the first feature information based on the observation grid (the average position of the point cloud and the first static point cloud number) and the second feature information based on the prediction grid (the predicted position coordinates and the second static point cloud number) through a specific association algorithm and screening rules. For example, in the grid association stage, it can be determined whether the predicted position of the prediction grid matches the position of the observation grid within a certain range, and the static point cloud numbers of the two are compared to see whether they meet the preset association conditions. If the predicted position of the prediction grid is close to the average position of the point cloud of a certain observation grid, and the static point cloud number relationship between the two meets the requirements, then the prediction grid may be regarded as a candidate for successful association with the observation grid.

[0025] Furthermore, in some examples, once the prediction grid is successfully associated with the observation grid, the prediction grid can be updated according to the information of the observation grid. The update process may involve, for example, the correction of the position coordinates, the adjustment of the number of static point clouds, and the update of other related attributes. By fusing the observation information and the prediction information in this way, a more accurate target prediction grid state can be obtained.

[0026] Further, in some examples, in the process of determining the target prediction grid, the prediction grid can be evaluated to determine whether it has the value of continuing tracking. Some prediction grids may be deleted because they are determined to have no tracking value due to various reasons. For example, if a prediction grid fails to associate with the observation grid after multiple attempts, and its lost frames exceed the preset extrapolation frame limit, then it can be considered that the possible target corresponding to the prediction grid has exceeded the trackable range or is a false prediction, and such a prediction grid will be deleted and no longer included in the scope of the target prediction grid. It can be seen that in this embodiment, deleting the prediction grid that has no tracking value helps to improve the efficiency and accuracy of the entire grid tracking system. It can reduce unnecessary computing resources wasted on those prediction grids that have lost their meaning, and avoid these invalid prediction grids from interfering with subsequent environmental perception and target judgment. The system can focus more on those target prediction grids that are truly valuable and correspond to objects or targets in the actual environment, so as to better achieve continuous and accurate tracking and status updates of objects in the environment, and provide more reliable support for grid-based applications (such as environmental perception and decision-making in autonomous driving).

[0027] For step S103, specifically, in some examples, the target prediction grid carries the estimated information of the location and related status of potential objects or targets in the environment. Each target prediction grid has its corresponding predicted position coordinates, number of static point clouds and other key attributes, which can reflect to a certain extent the approximate location and status characteristics of objects that may exist in the environment (such as other vehicles on the road, static obstacles, etc.) at the next moment.

[0028] Furthermore, in some examples, the predicted position coordinates of the target prediction grid can be associated and matched with the coordinates of the occupied grid in space. Since they are all constructed based on a unified grid map system, it is possible to determine which occupied grids or nearby areas of which occupied grids the target prediction grid corresponds to by means of coordinate comparison or the like. For example, if a target prediction grid predicts that there is a potential target at a certain coordinate position, then the corresponding or adjacent occupied grids can be found in the occupied grid map, so as to pay attention to the state changes such as the occupation probability of these occupied grids, and judge whether the object in the actual environment really appears in the area and whether its occupation state is as expected.

[0029] It is not difficult to find that compared with the related art, in the solution provided by the embodiment of the present application, since the prediction grids that have no tracking value have been deleted from the target prediction grid, there is no need to maintain the entire grid map, which reduces the high and time-consuming computational complexity and improves the tracking efficiency of the occupied grid. By tracking the occupied grid based on the target prediction grid, the entire system can achieve continuous monitoring of the position of static objects. It can be applied to timely discover obstacles on the road and reasonably plan driving routes in autonomous driving scenarios, which helps to improve the accuracy of the system's environmental perception and the rationality of decision-making.

[0030] Second embodiment

[0031] The second embodiment of the present application relates to a method for tracking an occupied grid. The second embodiment is an improvement on the first embodiment, and the specific improvement is that: in the second embodiment of the present application, a specific implementation method for determining the first characteristic information of the observation grid is provided.

[0032] Specifically, in some embodiments, the method for determining the first characteristic information based on the observation grid may include the following steps:

[0033] Step 201, determining the observation grid according to the point cloud and the grid map;

[0034] Step 202: extract first feature information for each observation grid.

[0035] Optionally, in some embodiments, the point cloud may be acquired by a sensor, for example, the sensor may be, but is not limited to, a 4D millimeter wave radar.

[0036] Specifically, in some examples, the point cloud can be rasterized to determine the characteristic information of the observation grid. Rasterization of the point cloud is a process of converting a point cloud (usually a set of discrete points in a three-dimensional space) into a grid map. The specific operation may be to first input a point cloud, and then project the point cloud into a grid map to form an observation grid. This is similar to "flattening" or "distributing" points in a three-dimensional space to a two-dimensional grid plane. For example, in an autonomous driving scenario, the environment around the vehicle can be obtained by a laser radar as a point cloud, and the point cloud is divided into a grid map centered on the vehicle through a certain algorithm. The grid map is a map formed by dividing the entire environmental space, which is a macroscopic concept. The grid map can be imagined as a large puzzle composed of many small squares (grids), which represents the complete range of the environment that the sensor can perceive.

[0037] Furthermore, in some examples, feature information can be extracted for each observation grid. The observation grid is formed in the process of rasterizing the point cloud, and is used to represent the unit after the spatial range that the sensor can observe is divided. It is a discretized representation of the environmental space, just like dividing the entire observable space into small squares, each of which is an observation grid. In other words, the observation grid is the basic unit that makes up the grid map, and is a local concept. Each observation grid is a small square in the grid map, like a small piece of a puzzle. For example, a grid map constructed around an autonomous vehicle may contain hundreds or thousands of observation grids, which together constitute a representation of the environment around the vehicle.

[0038] Furthermore, in some examples, the feature information may include the average position of the point cloud within the observation grid. and the number of static point clouds in the observation grid .

[0039] Specifically, for each observation grid, multiple point clouds may be projected, so it is necessary to comprehensively describe the location information of these point clouds. For example, suppose there are three point clouds in an observation grid, and the location coordinates of the three point clouds are: , and , then the average position of the point cloud in the observation grid The calculation formula can be:

[0040]

[0041] The average position of the point cloud in the observation grid makes it easier to understand the position where the point cloud in the corresponding observation grid is roughly concentrated.

[0042] Specifically, the static point cloud can be a static object such as a roadside building. In this embodiment, by counting the number of static point clouds in each observation grid, it is helpful to have a clearer understanding of the distribution of static objects in the environment. For example, if there are 5 static point clouds in an observation grid, the number "5" is the number of static point clouds in the observation grid, that is, .

[0043] It is not difficult to find that in the embodiment of the present application, a specific implementation method for determining the characteristic information of the observation grid is provided.

[0044] Third embodiment

[0045] The third embodiment of the present application relates to a method for tracking an occupied grid. The third embodiment is an improvement on the first embodiment. The specific improvement is that: in the third embodiment of the present application, a specific implementation method for determining the second feature information of the predicted grid is provided.

[0046] Specifically, in some embodiments, the method for determining the second feature information based on the prediction grid may include the following steps:

[0047] Step S301, determining a set of grid cells to be predicted; each grid cell corresponds to its own state information; the state information at least includes: the position coordinates of the grid cell in the previous frame, the number of the second static point cloud, the number of detection frames and the occupancy probability;

[0048] Step S302, determining predicted position coordinates according to the position coordinates of each grid unit in the previous frame;

[0049] Step S303: determining second feature information of the predicted grid according to the predicted position coordinates, the number of the second static point clouds, the number of detection frames and the occupancy probability.

[0050] Specifically, in some examples, the state information may include the position coordinates of the grid unit in the previous frame. , the number of the second static point cloud , detection frame number and occupation probability The detection frame number is used to represent the number of times the grid unit is detected, and the occupation probability is used to represent the possibility of the grid unit being occupied.

[0051] It can be understood that since the grid unit itself is stationary relative to the ground, it means that the grid unit itself will not actively change its position (excluding special extreme cases such as ground deformation). When making predictions, other state information except the position coordinates, such as the number of second static point clouds, the number of detection frames and the occupancy probability, remain unchanged. Only the position coordinates of the grid unit will be predicted to change accordingly based on the vehicle's own motion state.

[0052] Optionally, in some embodiments, the step of determining the predicted position coordinates according to the position coordinates of each grid unit in the previous frame, that is, step S302, may further include the following steps:

[0053] Step S3021, obtaining the sine value and cosine value of the frame-to-frame variation value of the yaw angular velocity of the vehicle;

[0054] Step S3022, obtaining the coordinate position change value of the vehicle between frames;

[0055] Step S3023, determining the predicted position coordinates according to the position coordinates of the previous frame, the coordinate position change value, the sine value and the cosine value.

[0056] Optionally, in some embodiments, the step S3023 of determining the predicted position coordinates according to the position coordinates of the previous frame, the coordinate position change value, the sine value and the cosine value may be specifically implemented by the following formula:

[0057] ;

[0058] ;

[0059] ;

[0060] in, represents the predicted position coordinates, represents the position coordinates of the grid unit in the previous frame, represents the coordinate position change value, represents the sine value, represents the cosine value.

[0061] The yaw rate is used to reflect the speed and direction of the vehicle's rotation around the vertical axis. In this embodiment, by incorporating the sine and cosine values ​​of the inter-frame change value of the yaw rate into the position calculation, the position change of the grid unit can be accurately calculated according to the vehicle's steering and other rotation conditions. The coordinate position change value, that is, the displacement of the vehicle in the horizontal direction (x-axis, y-axis direction) between the previous and next two frames, plays a key role in determining the position change of the grid unit relative to the vehicle.

[0062] For example, in the above formula, and From the calculation formula, it can be seen that this embodiment uses a combination of vector operations and trigonometric functions to comprehensively obtain the predicted and It can be understood that the yaw and displacement of the vehicle will change the relative position of the original grid unit in the direction. Since the horizontal movement of the vehicle (yaw, translation, etc.) has little effect on the vertical position of the grid unit, the z coordinate of the grid unit maintains the value of the previous frame, that is, .

[0063] It should be noted that this embodiment may also be an improvement based on the second embodiment.

[0064] It is not difficult to find that in an embodiment of the present application, a method for determining the second feature information based on the predicted grid is provided. This facilitates better use of grid information for environmental perception, path planning and other operations in subsequent application scenarios such as autonomous driving.

[0065] Fourth embodiment

[0066] The fourth embodiment of the present application relates to a method for tracking an occupied grid. The fourth embodiment is an improvement on the first embodiment, and the specific improvement is: in the fourth embodiment of the present application, a specific implementation method of grid association is provided.

[0067] Specifically, in some embodiments, determining the updated target prediction grid according to the first feature information and the second feature information, that is, step S102 may further include:

[0068] Step S1021, determining the association result between the observation grid and the prediction grid according to the first feature information and the second feature information;

[0069] Step S1022: determining an updated target prediction grid according to the association result.

[0070] In some embodiments, determining the association result between the observation grid and the prediction grid according to the first feature information and the second feature information, that is, step S1021 may include:

[0071] Step S10211, determining whether the predicted grid is within the grid map according to the predicted position coordinates;

[0072] Step S10212, if the predicted grid is within the grid map, determining the target observation grid directly corresponding to the predicted grid in the grid map;

[0073] Step S10213, determining the association result between the prediction grid and the target observation grid according to the first static point cloud quantity of the target observation grid and a preset threshold.

[0074] Specifically, in some examples, the positional relationship between the predicted position coordinates and the grid map can be determined, that is, whether the predicted position coordinates corresponding to the predicted grid fall within the range of the grid map. If the predicted position coordinates corresponding to the predicted grid fall within the range of the grid map, it can be determined that the predicted grid is within the grid map, thereby determining the target observation grid directly corresponding to the predicted grid in the grid map. Furthermore, the first static point cloud quantity of the target observation grid can be obtained. Assume that the preset threshold is Indicates that, according to the first static point cloud number of the target observation grid and the preset threshold , determine the association result between the prediction grid and the target observation grid.

[0075] Exemplarily, the preset threshold may be an empirical value, and relevant personnel may flexibly set it according to actual needs, and this embodiment does not make any specific limitation.

[0076] Optionally, in some embodiments, the determining the association result between the prediction grid and the target observation grid according to the number of the first static point cloud of the target observation grid and a preset threshold, that is, step S10213 may further include the following steps:

[0077] Step S102131, determining whether the number of the first static point cloud of the target observation grid is greater than the preset threshold;

[0078] Step S102132: if the number of the first static point cloud of the target observation grid is greater than the preset threshold, establishing an association relationship between the prediction grid and the target observation grid;

[0079] Step S102133, determining observation data according to the first characteristic information of the target observation grid; the observation data is used to update the prediction grid to determine an updated target prediction grid.

[0080] Specifically, in some examples, if the number of the first static point cloud is greater than the preset threshold, that is, , the prediction grid and the target observation grid can be associated, and the first feature information of the target observation grid can be extracted as observation data for updating the prediction grid.

[0081] That is to say, in this embodiment, if the prediction grid is suitable in position and the number of the first static point cloud of the target observation grid corresponding to the prediction grid also reaches a certain standard, then the prediction grid and the target observation grid can be associated, and the first feature information of the target observation grid can be extracted, and the first feature information can be retained as observation data. The prediction grid can be subsequently updated according to the observation data, so that the environmental information carried by the prediction grid can be adjusted in time with the new association situation.

[0082] In some embodiments, if the number of the first static point clouds is less than or equal to the preset threshold, the method may further include: determining a target adjacent observation grid based on the target observation grid; the target adjacent observation grid is used to characterize the grids adjacent to the target observation grid, in which the number of the first static point clouds is less than or equal to the preset threshold; and determining the association result between the prediction grid and the target observation grid based on the target observation grid and the target adjacent observation grid.

[0083] Specifically, in some examples, if the number of the first static point clouds is less than or equal to the set threshold, that is, , it means that the number of the first static point clouds at the location of the predicted grid does not meet the standard for directly associating the predicted grid and the target observation grid. At this time, the target adjacent observation grid can be determined, and then the association result of the predicted grid and the target observation grid can be determined based on the target observation grid and the target adjacent observation grid. Among them, the target adjacent observation grid is used to characterize the grid whose number of the first static point clouds is less than or equal to the preset threshold among the adjacent grids to the target observation grid. It can be understood that the target observation grid and the adjacent grids around it can be vividly represented by a nine-square grid. Usually, in a two-dimensional plane, the target observation grid is taken as the center, and a circle of eight adjacent grids around it plus the target observation grid itself form a total of nine grids in the form of a nine-square grid.

[0084] In some embodiments, determining the association result of the prediction grid and the target observation grid based on the target observation grid and the target adjacent observation grid may include: determining the sum of the first static point cloud number of the target observation grid and the first static point cloud number of the target adjacent observation grid; determining the association result of the prediction grid and the target observation grid based on the sum and a preset total threshold.

[0085] Specifically, in some examples, the sum of the number of the first static point cloud of the target observation grid and the number of the first static point cloud of the target adjacent observation grid can be calculated. Exemplarily, the sum can be expressed as The preset total threshold can be used express.

[0086] The preset total threshold may be an empirical value, and relevant personnel may flexibly set it according to actual needs, and this embodiment does not make any specific limitation.

[0087] In some embodiments, determining the association result of the prediction grid and the target observation grid based on the sum result and a preset total threshold may include: if the sum result is greater than the preset total threshold, associating the prediction grid with the target observation grid; determining a joint feature based on the first feature information of the target observation grid and the first feature information of the target adjacent observation grid; and determining the observation data based on the joint feature.

[0088] Specifically, in some examples, if the sum is greater than the preset total threshold, , then the prediction grid is associated with the target observation grid. Note that the prediction grid is not associated with the target adjacent observation grid. In this embodiment, the joint feature is used as the observation data. Since the joint feature can include the first feature information of the target observation grid and the first feature information of the target adjacent observation grid, it is possible to comprehensively consider the grid information around the target observation grid to more comprehensively and accurately update the environmental conditions reflected by the prediction grid.

[0089] Furthermore, in some embodiments, if the sum of the static point clouds is less than or equal to the preset total threshold, the method further includes: not establishing an association relationship between the prediction grid and the target observation grid.

[0090] Specifically, in some examples, if the sum is less than or equal to the preset total threshold, that is, , then the prediction grid is not associated with the target observation grid, and the target observation grid can be marked as used. That is to say, if the sum result is less than or equal to the preset total threshold, it means that the number of the first static point clouds of the target observation grid and the target adjacent observation grid has not reached the level that can be associated with the prediction grid. In this case, the prediction grid is not associated with the target observation grid, but the target observation grid is marked as used. Among them, marking the target observation grid as used is used to characterize that the target observation grid has been processed, such as it may have been excluded. In this way, repeated processing of the same target observation grid can be avoided, and the efficiency of the algorithm can be improved. In practical applications, this means that in subsequent iterations or processing steps, the target observation grid will be skipped.

[0091] It should be noted that this embodiment may also be an improvement based on the second embodiment and / or the third embodiment.

[0092] It is not difficult to find that in the embodiment of the present application, a specific implementation method for determining the updated target prediction grid based on the first feature information and the second feature information is provided. The decision process of how to establish a connection between the prediction grid and the observation grid in the existing grid map is disclosed, so as to extract appropriate observation data for subsequent prediction grid updates, ensure accurate perception of the environment and related systems (such as autonomous driving systems) to make effective decisions and operations based on accurate grid information.

[0093] Fifth embodiment

[0094] The fifth embodiment of the present application relates to a method for tracking an occupied grid. The fifth embodiment is an improvement on the fourth embodiment, and the specific improvement is that: in the fifth embodiment of the present application, a specific implementation method for grid update is provided.

[0095] Specifically, in some embodiments, the method of determining an updated target prediction grid according to the association result, that is, step S1022, may further include the following steps:

[0096] According to the observation data, the target attribute of the prediction grid is updated; the target attribute includes at least one of the following: the predicted position coordinates and the accumulated static point cloud number of the prediction grid, and the accumulated static point cloud number is used to represent the accumulated value of the second static point cloud number.

[0097] Specifically, in some examples, after the prediction grid is successfully associated with the target observation grid and the observation data is obtained, the newly obtained observation data can be used to optimize and update the state of the prediction grid. Exemplarily, the Kalman filter algorithm, a commonly used method that can effectively integrate prediction information and observation information, can be used with the help of the observation data to make the state of the prediction grid in multiple aspects more in line with the actual situation, including target attributes such as position coordinates and accumulated static point cloud numbers, and other related attributes of the prediction grid can also be updated.

[0098] In some embodiments, the target attribute is specifically the predicted position coordinates of the prediction grid; and updating the target attribute of the prediction grid according to the observation data may include: step S401, determining the average position of the point cloud of the target observation grid according to the observation data;

[0099] Step S402: updating the predicted position coordinates of the prediction grid according to the average position of the point cloud, the predicted position coordinates of the prediction grid and the Kalman filter gain.

[0100] It can be understood that in this embodiment, in combination with the Kalman filter algorithm, a better estimated value is obtained based on the current coordinate value (prediction value) of the prediction grid and the observation value corresponding to the observation data to implement the update operation of the predicted position coordinates of the prediction grid.

[0101] Optionally, in some embodiments, the predicted position coordinates of the prediction grid are updated according to the average position of the point cloud, the predicted position coordinates of the prediction grid and the Kalman filter gain, that is, step S402 can be specifically implemented by the following formula:

[0102] ;

[0103] ;

[0104] ;

[0105] in, represents the current position coordinates of the prediction grid after updating, is the predicted position coordinate of the prediction grid, is the average position of the point cloud, represents the Kalman filter gain.

[0106] In some examples, assume that the current x coordinate of the prediction grid is , Represents the x-coordinate of the prediction grid obtained by the previous prediction; is the observed value of the x-coordinate obtained through the observation data. is the Kalman filter gain, which plays a key role in weighing the weights of the predicted value and the observed value in the entire updating process. When the values ​​are different, the final updated In this embodiment, based on the x-coordinate of the prediction grid, the deviation between the observed value and the predicted value is , multiplied by the Kalman filter gain After that, the predicted value is corrected to obtain a more accurate x-coordinate of the predicted grid for the current frame. and The coordinate update formula also follows the same principle, using the deviation between the corresponding observed value and the predicted value, and combining the respective Kalman filter gains to correct the predicted y-coordinate and z-coordinate to obtain a more accurate grid position coordinate of the current frame.

[0107] It should be noted that the Kalman filter gain The settings can be adjusted according to different application scenarios and accuracy requirements, and the settings can be pre-configured in the system or dynamically adjusted according to real-time requirements during the operation of the algorithm to optimize the accuracy and stability of the prediction grid update. This embodiment does not set the Kalman filter gain. The value of is limited.

[0108] Optionally, in some embodiments, the target attribute is specifically the cumulative number of static point clouds; updating the target attribute of the prediction grid according to the observation data may include: step S501, determining the first static point cloud number of the target observation grid according to the observation data; step S502, updating the target attribute of the prediction grid according to the first static point cloud number and the cumulative static point cloud number of the prediction grid in the previous frame.

[0109] Optionally, in some embodiments, the target attribute of the prediction grid is updated according to the first static point cloud quantity and the accumulated static point cloud quantity of the prediction grid in the previous frame, that is, step S502 can be specifically implemented by the following formula:

[0110] ;

[0111] Among them, the represents the cumulative static point cloud number of the prediction grid after the prediction grid is updated in the current frame, represents the number of accumulated static point clouds of the predicted grid in the previous frame, Represents the number of the first static point cloud of the target observation grid.

[0112] Specifically, in some examples, the cumulative static point cloud number of the prediction grid in the previous frame is , that is, the total number of static point clouds that have been counted in the prediction grid as of the last frame. And the number of the first static point cloud of the target observation grid , is the number of static point clouds contained in the first feature information of the target observation grid associated with the prediction grid this time; further, by adding the cumulative number of static point clouds of the previous frame of the prediction grid and the newly acquired number of static point clouds of the target observation grid, the cumulative number of static point clouds of the prediction grid after the current frame is updated can be obtained, so that the continuous tracking of the changes in the number of static point clouds in the prediction grid can be achieved.

[0113] Optionally, in some embodiments, after determining the observation data, the method further includes: updating relevant attributes in the prediction grid except the target attribute; the relevant attributes include at least one of the following: a maturity flag, a number of detected frames, and a number of lost frames.

[0114] Specifically, in some examples, in addition to updating the target attributes of the prediction grid as described above, some other related attributes may also be updated in this embodiment, and the other related attributes may include but are not limited to at least one of the following: a maturity flag, a number of detected frames, and a number of lost frames.

[0115] Among them, the maturity flag is used to indicate whether the relevant information of the prediction grid is sufficiently stable and reliable and has reached a state that can be considered "mature". The update of the maturity flag can be comprehensively judged based on the observation data associated with this time and the situation of the update operation.

[0116] Among them, the number of detected frames and the number of lost frames are related to the number of times the prediction grid is detected or not detected at different stages. With each new association and the advancement of the frame, the number of detected frames and the number of lost frames need to be adjusted accordingly to more comprehensively and accurately reflect the state changes of the prediction grid.

[0117] Exemplarily, the update operation of the detection frame number can be implemented based on the detection frame number of the prediction grid in the previous frame. For example, the update operation of the detection frame number can be implemented by using the following formula:

[0118] ;

[0119] Among them, the represents the number of detection frames of the prediction grid after the current frame is updated, Indicates the detection frame number of the prediction grid in the previous frame.

[0120] It can be understood that after each frame of detection, the number of detection frames should naturally increase by 1, so the number of detection frames of the prediction grid in the current frame is updated by adding 1 to the number of detection frames of the previous frame based on the prediction grid, so as to record the total number of times the prediction grid is detected.

[0121] It should be noted that this embodiment may also be an improvement based on any one or more of the first to third embodiments.

[0122] It is not difficult to find that in the embodiment of the present application, a specific implementation method for determining the updated target prediction grid according to the association result is provided. For example, in this embodiment, by reasonably utilizing observation data and Kalman filter algorithm and other means, multiple target attributes of the prediction grid are updated, so that the environmental information carried by the prediction can be continuously optimized and improved with new observations, providing a more accurate data basis for subsequent environmental perception, decision-making and other operations based on grid information.

[0123] Sixth embodiment

[0124] The sixth embodiment of the present application relates to a method for tracking an occupied grid. The sixth embodiment is an improvement based on the fourth embodiment, and the specific improvement is: in the sixth embodiment of the present application, a specific implementation method for determining an updated target prediction grid according to the association result is provided.

[0125] Specifically, in some embodiments, determining the updated target prediction grid according to the association result, that is, step S1022, may also include the following steps:

[0126] Step S10221, determining the prediction grid in the association result that has not established an association relationship with the target observation grid;

[0127] Step S10222, determining the number of extrapolated frames according to the type of region where the prediction grid is located and the maturity state of the prediction grid;

[0128] Step S10223, performing an extrapolation operation according to the extrapolated frame number.

[0129] Specifically, in some examples, grid extrapolation is mainly used to process prediction grids that have not established an association relationship with the target observation grid. Since these prediction grids do not have new observation data to update the status of the prediction grid, it is necessary to use the extrapolation method to infer the possible subsequent situation of the prediction grid based on certain rules. At the same time, it is also necessary to determine the number of extrapolated frames based on different areas (such as blind spots and non-blind spots) and the maturity status of the prediction grid, and determine when the prediction grid is no longer worth tracking. Then, it can be determined whether to delete the prediction grid, so as to reasonably manage and update the prediction grid and ensure the effectiveness and accuracy of the entire environmental perception system based on the prediction grid.

[0130] The mature state may include a mature state and an immature state. If the predicted grid is in a mature state, the predicted grid is a mature grid. If the predicted grid is in an immature state, the predicted grid is an immature grid. The mature condition of the mature grid includes but is not limited to the cumulative static point cloud number reaching a certain threshold, the number of detection frames exceeding a specific value, or based on other relevant feature information combination judgment, etc., and the mature condition can be adjusted according to different application scenarios and needs. The adjustment method is that it can be pre-configured in the system or dynamically adjusted according to real-time needs during the operation of the algorithm to adapt to the requirements for grid maturity judgment in different environments, and ensure the coordination of the grid extrapolation operation with the entire environment perception and target tracking algorithm.

[0131] In some embodiments, the extrapolation frame number is determined according to the area type where the prediction grid is located and the maturity state of the prediction grid, that is, step S10222 may include: if the area type where the prediction grid is located is a blind spot, determining whether the blind spot grid is a mature grid; wherein the blind spot grid is used to characterize the prediction grid located in the blind spot; if the blind spot grid is a mature grid, obtaining the current extrapolation frame number; performing the extrapolation operation according to the extrapolation frame number is specifically: performing an extrapolation operation on the blind spot grid according to the current extrapolation frame number; if the blind spot grid is an immature grid, obtaining a preset first frame number; performing the extrapolation operation according to the extrapolation frame number is specifically: performing an extrapolation operation on the blind spot grid according to the first frame number.

[0132] In some examples, for the prediction grid in the blind area, i.e., the blind area grid, it can be determined whether the blind area grid is a mature grid. If the blind area grid is a mature grid, it can continue in the established extrapolation mode. In this way, even if there is no new observation data associated, the extrapolation state of the mature grid is still retained, and the subsequent situation of the mature grid is inferred, because the information accumulated by such mature grids is relatively reliable and has the value of continuing extrapolation.

[0133] Furthermore, in some examples, if the blind area grid is an immature grid, the situation is different, and a preset first frame number a can be obtained, and an extrapolation operation is performed on the blind area grid according to the first frame number to update the state or position of the prediction grid. The preset first frame number a is used to characterize the number of frames that are not associated with the observation data, so as to limit the number of extrapolated frames for the prediction grid in the immature blind area.

[0134] Optionally, in some embodiments, the extrapolating operation on the blind spot grid according to the first frame number may include: determining the first lost frame number of the blind spot grid being extrapolated; and deleting the corresponding predicted grid according to the first lost frame number and the first frame number so that the predicted grid no longer participates in the subsequent tracking process.

[0135] In some examples, the first number of lost frames is used to characterize the number of times the predicted grid loses tracking. As long as the first number of lost frames is less than the first number of frames a, the blind area grid will continue to be extrapolated; however, once the first number of lost frames is greater than or equal to the first number of frames a, it indicates to a certain extent that the corresponding blind area grid may have lost the meaning of continuing to track, so the corresponding blind area grid may be deleted and no further tracking of the blind area grid is performed.

[0136] In some embodiments, determining the number of extrapolated frames according to the area type where the predicted grid is located and the maturity state of the predicted grid may also include the following steps: if the area type where the predicted grid is located is a non-blind area, determining whether the non-blind area grid is a mature grid; wherein the non-blind area grid is used to characterize the predicted grid located in the non-blind area; if the non-blind area grid is a mature grid, obtaining a preset second number of frames; performing an extrapolation operation according to the number of extrapolated frames is specifically: performing an extrapolation operation on the non-blind area grid according to the second number of frames; if the non-blind area grid is an immature grid, obtaining a preset third number of frames; performing an extrapolation operation according to the number of extrapolated frames is specifically: performing an extrapolation operation on the non-blind area grid according to the third number of frames; the third number of frames is less than the second number of frames.

[0137] In some examples, for a mature grid in a non-blind area, a preset second frame number b may be obtained, and an extrapolation operation may be performed on the non-blind area grid according to the second frame number b; for an immature grid in a non-blind area, a preset third frame number c may be obtained, and an extrapolation operation may be performed on the non-blind area grid according to the third frame number c. The third frame number c is less than the second frame number b.

[0138] In some embodiments, the extrapolating operation on the non-blind area grid according to the second frame number may include: determining the second lost frame number of the non-blind area grid being extrapolated; deleting the corresponding predicted grid according to the second lost frame number and the second frame number, so that the predicted grid no longer participates in the subsequent tracking process; the extrapolating operation on the non-blind area grid according to the preset third frame number includes: determining the third lost frame number of the non-blind area grid being extrapolated; deleting the corresponding predicted grid according to the third lost frame number and the third frame number, so that the predicted grid no longer participates in the subsequent tracking process.

[0139] Specifically, in some examples, as long as the second number of lost frames is less than the second number of frames b, the non-blind grid will continue to be extrapolated, and the extrapolation operation will be used to maintain the speculation of the non-blind grid situation. When the second number of lost frames is greater than or equal to the second number of frames b, it means that the non-blind grid has no new observation data associated for a long time, and it is no longer necessary to continue tracking. Therefore, the non-blind grid can be deleted. Similarly, as long as the third number of lost frames is less than the third number of lost frames c, the non-blind grid will continue to be extrapolated, and the extrapolation operation will be used to maintain the speculation of the non-blind grid situation. When the third number of lost frames is greater than or equal to the third number of frames c, it means that the non-blind grid has no new observation data associated for a long time, and it is no longer necessary to continue tracking. Therefore, the non-blind grid can be deleted.

[0140] Among them, the first frame number a, the second frame number b and the third frame number c are all adjustable parameters, which can be set according to different application scenarios, environmental conditions and accuracy requirements. The setting method includes pre-configuration during system initialization or dynamic adjustment according to real-time monitoring data and algorithm feedback during system operation to optimize the adaptability and effectiveness of the prediction grid extrapolation operation, which is conducive to further improving the overall performance of the environmental perception system.

[0141] It should be noted that this embodiment may also be an improvement based on any one or more of the first to third embodiments and the fifth embodiment.

[0142] It is not difficult to find that in the embodiment of the present application, a method for updating the prediction grid that has not established an association relationship with the target observation grid is provided to obtain an updated target prediction grid. Through this embodiment, different regions and factors such as the maturity status of the prediction grid can be carefully distinguished, and corresponding extrapolation frame number rules and grid deletion conditions are formulated. In the absence of observation data association, the subsequent situation of the prediction grid is speculated as reasonably as possible, and the prediction grid that has lost its tracking value is cleaned up in time, so that the entire grid system can efficiently and accurately reflect changes in environmental status, and provide reliable data support for applications such as environmental perception and path planning in scenarios such as autonomous driving.

[0143] Seventh embodiment

[0144] The seventh embodiment of the present application relates to a method for tracking an occupied grid. The seventh embodiment is an improvement on the fourth embodiment, and the specific improvement is: in the seventh embodiment of the present application, another specific implementation method for determining an updated target prediction grid based on the association result is provided, especially whether a new grid is obtained.

[0145] Specifically, in some embodiments, determining the updated target prediction grid according to the association result, that is, step S1022, may further include the following steps:

[0146] Step S1022a, determining an idle observation grid in the current frame; the idle observation grid is used to represent an unused observation grid in the grid map;

[0147] Step S1022b, combining the number of first static point clouds in the idle observation grid and the preset threshold, determining whether to obtain a new grid based on the idle observation grid.

[0148] Specifically, in some examples, obtaining a new grid based on the idle observation grid refers to a process of determining which observation grids can be added to the grid list as new elements in the grid map of the current frame. This is very important for timely updating and expanding the environmental information covered by the grid map, allowing the system to continuously incorporate new observation grids that meet certain conditions, thereby more comprehensively and accurately reflecting the actual environmental conditions, and providing a richer data foundation for subsequent various applications based on observation grids (such as environmental perception and path planning in autonomous driving).

[0149] Specifically, in some examples, observation grids that have not been used in the current frame grid map, i.e., idle observation grids, can be first determined to perform conditional judgment on the idle observation grids. One of the judgment conditions is to combine the number of first static point clouds in the idle observation grids and the preset threshold to determine whether to obtain a new grid based on the idle observation grids.

[0150] In some embodiments, the combination of the first static point cloud number in the idle observation grid and the preset threshold value determines whether to obtain a new grid based on the idle observation grid, that is, step S1022b may include: if the first static point cloud number in the idle observation grid is greater than the preset threshold value, then obtaining a new grid based on the idle observation grid; if the first static point cloud number in the idle observation grid is less than or equal to the preset threshold value, then determining whether to obtain a new grid based on the idle observation grid based on the sum of the first static point cloud number and the preset total threshold value.

[0151] Specifically, in some examples, when the number of the first static point clouds in the idle observation grid is greater than the preset threshold, that is, When determining that the corresponding idle observation grid has reached a certain standard in terms of the number of the first static point cloud, and can provide relatively reliable and valuable environmental information, it can be regenerated, and the corresponding idle observation grid is added to the grid list as a new grid, so that the new grid becomes a part of the system that can be used for subsequent operations and analysis. Among them, the regeneration operation can be performed in accordance with the pre-set grid regeneration rules, and the grid regeneration rules can include but are not limited to: initialization settings of the new grid, positioning in the grid map, and related attribute assignment operations, so as to ensure that the new grid can be normally integrated into the subsequent environmental perception and data processing process.

[0152] Furthermore, when the number of the first static point clouds in the idle observation grid is less than or equal to the preset threshold, it cannot be immediately determined whether the idle observation grid can be newly generated, but it is necessary to further investigate the surrounding conditions of the idle observation grid.

[0153] Optionally, in some embodiments, determining whether the corresponding idle observation grid can be newly generated based on the sum of the first static point cloud numbers and the preset total threshold may include: if the sum of the first static point cloud numbers is greater than the preset total threshold, obtaining a newly generated grid based on the idle observation grid; if the sum of the first static point cloud numbers is less than or equal to the preset total threshold, not obtaining a newly generated grid based on the idle observation grid.

[0154] In some examples, the sum of the number of the first static point clouds in the nine-square grid around the idle observation grid (that is, the 3×3 area consisting of 8 observation grids around the idle observation grid plus the idle observation grid itself) can be calculated. If the sum of the first static point clouds is greater than the preset total threshold, Even if the number of first static point clouds of the idle observation grid itself is not enough, if the sum of the number of first static point clouds of the surrounding area meets certain requirements, the corresponding idle observation grid still has the qualification of being reborn and can be added to the grid list to participate in the subsequent processing flow as a new grid.

[0155] Further, in some examples, if the sum of the first static point cloud quantities is less than or equal to the preset total threshold, that is, , it means that the number of static point clouds of the idle observation grid itself and the area around the idle observation grid as a whole has not reached the standard for the idle observation grid to be regenerated, so the idle observation grid will not be regenerated and will not be included in the grid list.

[0156] It should be noted that this embodiment may also be an improvement based on any one or more of the first to third embodiments, the fifth embodiment and the sixth embodiment.

[0157] It is not difficult to find that in the embodiment of the present application, a rule for obtaining a new grid is provided, which discloses how to judge whether an unused observation grid in a grid map is qualified to be added to the system as a new grid, and by comprehensively considering the number of first static point clouds of the observation grid itself and its surrounding area, checking the number of first static point clouds of a single observation grid and the sum of the number of first static point clouds including the observation grid, it is determined whether to mark the corresponding observation grid as a new grid. It can ensure that the newly included new grid can bring valuable supplements to the environmental information reflected by the entire grid system, ensure the quality and relevance of the grid in the grid list, and thus better serve the requirements of relevant application scenarios for environmental perception accuracy.

[0158] Eighth embodiment

[0159] The eighth embodiment of the present application relates to a method for tracking an occupied grid. The eighth embodiment is an improvement on the first embodiment, and the specific improvement is: in the eighth embodiment of the present application, a specific implementation method for grid post-processing of tracking the occupied grid according to the target prediction grid is provided.

[0160] Specifically, in some embodiments, the tracking of the occupied grid is implemented according to the target predicted grid, that is, step S103 may further include the following steps:

[0161] Step S1031, obtaining the cumulative static point cloud number and the detection frame number of each target prediction grid; the cumulative static point cloud number is used to represent the cumulative value of the second static point cloud number;

[0162] Step S1032, determining the occupancy probability of the target prediction grid according to the accumulated static point cloud number and the detection frame number;

[0163] Step S1033: Tracking the occupied grid according to the occupancy probability of the target predicted grid.

[0164] In some embodiments, the occupancy probability of the target prediction grid is determined according to the accumulated static point cloud number and the detection frame number, that is, step S1032 can be specifically implemented by the following formula:

[0165]

[0166] in, represents the occupation probability, represents the number of accumulated static point clouds, Indicates the detection frame number.

[0167] In some examples, the cumulative number of static point clouds reflects the cumulative value of the number of the second static point clouds accumulated in a target prediction grid over a period of time (after multiple frames of detection). The cumulative number of static point clouds can reflect the frequency or number of static objects appearing in the area corresponding to the target prediction grid; the number of detection frames is the total number of times the corresponding target prediction grid is detected.

[0168] In this embodiment, the possibility of the corresponding target prediction grid being occupied by the object is measured by subtracting 1 from the cumulative static point cloud number and dividing it by the detection frame number, taking into account the cumulative static point cloud number and the detection frame number in the target prediction grid.

[0169] Optionally, in some embodiments, the tracking of the occupied grid is implemented according to the occupancy probability of the target predicted grid, that is, step S1033 may further include the following steps:

[0170] Step S10331, deleting the target prediction grids whose occupancy probabilities are less than or equal to the occupancy probability threshold, to obtain the retained prediction grids;

[0171] Step S10332: Tracking the occupied grid according to the retained predicted grid.

[0172] In some examples, after calculating the occupancy probability of each of the target prediction grids, those target prediction grids with lower occupancy probabilities can be screened out based on the occupancy probability. In this embodiment, by setting the occupancy probability threshold, only target prediction grids with an occupancy probability higher than the occupancy probability threshold will be retained and displayed. It can be seen that this embodiment takes into account that the target prediction grids with low occupancy probabilities are likely to correspond to those areas where some point clouds occasionally appear or a small number of static point clouds are detected but are not stable and it is not certain that there are actual objects occupying the area. By deleting the corresponding target prediction grids, the retained prediction grids that are finally displayed can be more focused on those areas where there is a high probability that objects do exist and are more valuable to environmental perception, making the presented results more significant and facilitating more accurate judgments and decisions based on the retained prediction grids.

[0173] The occupancy probability threshold may be adjusted and set according to different application scenarios, display requirements, and accuracy requirements, and this embodiment does not specifically limit this.

[0174] For example, Figure 2 The figure shows the distribution of parking spaces in a certain park, where white dots indicate that the probability of a parking space being occupied is high, and each white dot represents a reserved prediction grid, that is, a parking space. The black area indicates an area that is not occupied by a parking space, which may be a non-parking area such as a road, or a parking area that is not shown in the figure because the probability of a parking space being occupied is low.

[0175] In this example, the number of retained prediction grids is about several hundred to one thousand. Figure 2 There may be overlaps in the white dots, but each white dot represents an independent retained prediction grid. In addition, in practical applications, relevant personnel can adjust the size of the white dots according to actual display requirements to provide clearer display information.

[0176] It should be noted that this embodiment may also be an improvement based on any one or more of the second to seventh embodiments.

[0177] It is not difficult to find that in the embodiment of the present application, a specific implementation method is provided for tracking the occupied grid according to the target prediction grid. For example, by obtaining the retained prediction grid, the target prediction grid is further optimized and purified, aiming to improve the quality and effectiveness of the grid information in reflecting the environmental conditions, and provide more reliable and critical environmental perception data for related application scenarios. In this way, it can be ensured that the grids finally presented are of significant significance to the overall environmental perception, avoiding excessive unimportant or interfering grid information affecting subsequent use.

[0178] The step division of the above methods is only for clear description. When implemented, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application; adding insignificant modifications to the algorithm or process or introducing insignificant designs without changing the core design of the algorithm and process are all within the scope of protection of this application.

[0179] In summary, this application has at least the following beneficial effects:

[0180] (1) The method for tracking occupied grids of the present application is particularly effective for tracking sparse occupied grids. The method is significantly different from the traditional method. The traditional method requires the maintenance of the entire grid map, involves a large amount of grid data processing, and has high computational complexity and is time-consuming. The algorithm provided in the present application only focuses on tracking grids that meet the point cloud quantity requirements. Through such a selective tracking method, it cleverly avoids processing all grids, fundamentally reduces the amount of data that needs to be calculated, and thus greatly reduces the overall complexity of the algorithm. It can also significantly shorten the time spent on calculations, thereby improving the efficiency of the algorithm in practical applications.

[0181] (2) The occupied grid tracking method of the present application predicts, associates, updates, generates, and post-processes the grid, and introduces the idea of ​​track management. By setting the number of frames required for the grid to mature and the number of frames for the grid to disappear, it can be ensured that the grid will not remain noticeably in the scene, and at the same time, too many false grids will not be generated, thereby improving the accuracy and efficiency of tracking.

[0182] (3) The occupancy grid tracking method of the present application can be implemented based on 4D millimeter wave radar. Compared with lidar, its advantages are that it has a longer detection range and can resist the impact of bad weather and work normally under bad weather conditions.

[0183] Ninth embodiment

[0184] The ninth embodiment of the present application also provides an electronic device. The electronic device can be various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device can also be various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices.

[0185] The electronic device includes: one or more processors; and a memory storing computer program instructions, wherein when the computer program instructions are executed, the processor executes the steps of the method provided in any one or more of the above embodiments. Figure 3 An exemplary structural diagram of the electronic device is disclosed. Figure 3 As shown, the electronic device includes: one or more processors 1101, a memory 1102, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, if necessary, multiple processors and / or multiple buses can be used with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Among them, the components shown in this article, their connections and relationships, and their functions are only examples, and are not intended to limit the implementation of the present application described and / or required herein.

[0186] The electronic device may further include: an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103 and the output device 1104 may be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0187] The input device 1103 can receive input digital or character information, and generate key signal input related to the user settings and function control of the electronic device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator rod, one or more mouse buttons, a trackball, a joystick and other input devices. The output device 1104 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display and a plasma display. In some embodiments, the display device may be a touch screen.

[0188] In order to provide interaction with the user, the electronic device may be a computer. The computer has: a display device (e.g., a cathode ray tube (CRT) or an LCD monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball), through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0189] In the embodiments of the present application, a computer program / instruction is stored on a computer-readable medium, and when the computer program / instruction is executed by a processor, the steps of the method provided by any one or more of the above embodiments are implemented. The computer-readable medium may be included in the electronic device described in the above embodiments; or it may exist independently without being assembled into the device. The above computer-readable medium carries one or more computer-readable instructions.

[0190] The memory 1102 can be used as a non-transient computer-readable storage medium, which can be used to store non-transient software programs, non-transient computer executable programs and modules. The processor 1101 executes various functional applications and data processing of the server by running the non-transient software programs, instructions and modules stored in the memory 1102, so as to implement the program instructions / modules corresponding to the method provided by any one or more embodiments in the embodiments of the present application.

[0191] The memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 1102 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 1102 may optionally include a memory remotely arranged relative to the processor 1101, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0192] It should be noted that the computer-readable medium described in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM, Erasable Programmable Read - Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM, Compact Disc Read - Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0193] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0194] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0195] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. For example, an application-specific integrated circuit (ASIC), a general-purpose computer or any other similar hardware device may be used for implementation. In some embodiments, the software program of the present application may be executed by a processor to implement the above steps or functions. Similarly, the software program of the present application (including related data structures) may be stored in a computer-readable recording medium, such as a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some steps or functions of the present application may be implemented by hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

[0196] The computer program product provided in the embodiment of the present application includes one or more computer programs / instructions, which, when executed by the processor, generate in whole or in part the process or function described in the embodiment of the present application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a server, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL, Digital Subscriber Line)) or wireless (e.g., infrared, wireless, microwave, etc.) means to another website site, computer, server, or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrated therein. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state drive, SSD, solid statedisk), etc.

[0197] The flow chart or block diagram in the accompanying drawings shows the possible architecture, function and operation of the equipment, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated system for hardware that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0198] The scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present application. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim may also be implemented by one unit or device through software or hardware. The words "first", "second", etc. are only used to distinguish the description, and do not indicate any particular order, nor can they be understood as indicating or implying relative importance.

[0199] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims, and the above embodiments should be regarded as exemplary and non-restrictive.

Claims

1. A method for tracking an occupied grid, characterized in that: The method comprises: Acquire first feature information based on the observation grid and second feature information based on the prediction grid; wherein the first feature information includes the average position of the point cloud in the observation grid and the number of the first static point cloud, and the second feature information includes the predicted position coordinates of the prediction grid and the number of the second static point cloud; According to the predicted position coordinates, determine whether the predicted grid is in the grid map; if the predicted grid is in the grid map, determine the target observation grid directly corresponding to the predicted grid in the grid map; according to the number of the first static point cloud of the target observation grid and a preset threshold, determine the association result between the predicted grid and the target observation grid; according to the association result, determine the updated target predicted grid; the predicted grid that has no tracking value has been deleted from the target predicted grid; Tracking the occupied grid according to the target prediction grid; Among them, if the first static point cloud quantity is less than or equal to the preset threshold, the method includes: determining a target adjacent observation grid according to the target observation grid; the target adjacent observation grid is used to characterize the grid whose first static point cloud quantity is less than or equal to the preset threshold among the adjacent grids to the target observation grid; determining the sum of the first static point cloud quantity of the target observation grid and the first static point cloud quantity of the target adjacent observation grid; if the sum is greater than the preset total threshold, associating the prediction grid with the target observation grid; determining a joint feature according to the first feature information of the target observation grid and the first feature information of the target adjacent observation grid; and determining observation data according to the joint feature.

2. The method according to claim 1, characterized in that The determining the association result between the prediction grid and the target observation grid according to the first static point cloud quantity of the target observation grid and a preset threshold comprises: Determine whether the number of the first static point cloud of the target observation grid is greater than a preset threshold; If the number of the first static point cloud of the target observation grid is greater than the preset threshold, establishing an association relationship between the prediction grid and the target observation grid; Observation data is determined according to the first characteristic information of the target observation grid; the observation data is used to update the prediction grid to determine an updated target prediction grid.

3. The method according to claim 1, characterized in that Determining an updated target prediction grid according to the association result includes: According to the observation data, the target attribute of the prediction grid is updated; the target attribute includes at least one of the following: the predicted position coordinates and the accumulated static point cloud number of the prediction grid; the accumulated static point cloud number is used to characterize the accumulated value of the second static point cloud number.

4. The method according to claim 3, characterized in that The target attribute is specifically the predicted position coordinates of the prediction grid; The updating of the target attribute of the prediction grid according to the observation data comprises: Determining the average position of the point cloud of the target observation grid according to the observation data; The predicted position coordinates of the prediction grid are updated according to the average position of the point cloud, the predicted position coordinates of the prediction grid and the Kalman filter gain.

5. The method according to claim 1, characterized in that: Determining an updated target prediction grid according to the association result includes: Determine, in the association result, the prediction grid that has not established an association relationship with the target observation grid; The number of extrapolated frames is determined according to the type of region where the prediction grid is located and the maturity state of the prediction grid; wherein, if the prediction grid is in a mature state, the prediction grid is a mature grid, and the maturity condition of the mature grid includes at least one of the following: the number of accumulated static point clouds reaches a certain threshold, and the number of detection frames exceeds a specific value; An extrapolation operation is performed according to the extrapolated frame number.

6. The method according to claim 5, characterized in that Determining the number of extrapolated frames according to the region type where the prediction grid is located and the maturity state of the prediction grid includes: If the area type where the prediction grid is located is a blind area, determining whether the blind area grid is a mature grid; wherein the blind area grid is used to characterize the prediction grid located in the blind area; If the blind area grid is a mature grid, the current extrapolation frame number is obtained; the performing the extrapolation operation according to the extrapolation frame number is specifically: performing the extrapolation operation on the blind area grid according to the current extrapolation frame number; If the blind area grid is an immature grid, a preset first frame number is obtained; and performing the extrapolation operation according to the extrapolated frame number is specifically: performing the extrapolation operation on the blind area grid according to the first frame number.

7. The method according to claim 1, characterized in that Determining an updated target prediction grid according to the association result includes: Determining an idle observation grid in the current frame; the idle observation grid is used to represent an unused observation grid in the grid map; In combination with the number of first static point clouds in the idle observation grid and the preset threshold, it is determined whether to obtain a new grid according to the idle observation grid.

8. An electronic device, characterized in that: The electronic device comprises: one or more processors; and A memory storing computer program instructions, which, when executed, cause the processor to perform the steps of the method as claimed in any one of claims 1 to 7.

9. A computer readable medium having a computer program / instructions stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images