Map Optimization Method, Device, Equipment and Storage Medium

Abstract

The present invention relates to the technical field of map making, and particularly relates to a map optimization method, device, equipment and storage medium. The method includes obtaining a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame; extracting point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determining whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing processing condition; if satisfied, determining a distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determining a corresponding weight value based on the distance value by using a preset weight function; performing a smoothing transition process on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map. The present application smooths two point cloud frames without feature continuity by increasing the weight value, so as to solve the problem of discontinuous point cloud frame features in the process of making a map by the existing normal distribution change algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of map making, and in particular, to a map optimization method, device, equipment and storage medium. Background Art

[0002] The high-precision map required for autonomous driving is made from the data collected by a vehicle equipped with a collection device, and a lidar is a commonly used collection device. When making a map from the data collected by the lidar, the normal distribution transformation (NDT) algorithm is usually used to perform the matching between point cloud frames. Since the NDT algorithm can be applied to the statistical model of three-dimensional points, this algorithm can better determine the optimal matching between two point clouds. However, this algorithm does not utilize the feature calculation and matching of corresponding points during the registration process. The map obtained by such a point cloud frame matching method will have the problem of discontinuous point cloud frame features. Summary of the Invention

[0003] The main purpose of the present invention is to solve the problem of discontinuous point cloud frame features existing in the process of making a map by the existing normal distribution change algorithm.

[0004] The first aspect of the present invention provides a map optimization method, and the map optimization method includes:

[0005] Obtain a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame, where the target point cloud frame is a point cloud frame to be optimized in the point cloud map, and the neighborhood grid includes at least one point cloud frame;

[0006] Extract the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determine whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing processing condition;

[0007] If satisfied, determine the 5 distance values between the target point cloud frame and the corresponding point cloud frames in the neighborhood grid, and determine the corresponding weight values based on the distance values by using a preset weight function;

[0008] Perform a smooth transition process on the target point cloud frame and the corresponding point cloud frames in the neighborhood grid based on the weight values to obtain an optimized point cloud map.

[0009] Optionally, in the first implementation manner of the first aspect of the present invention, the obtaining of the

[0010] target point cloud frame of the point cloud map and the neighborhood grid of the target point cloud frame includes: Select at least one target point cloud frame from the point cloud map;

[0011] Search for the point cloud grids around each target point cloud frame as the center, and calculate the overlap rate and point cloud probability density between the searched point cloud grids;

[0012] Determine the neighborhood grid of the target point cloud frame based on the overlap rate and the point cloud probability density.

[0013] Optionally, in the second implementation manner of the first aspect of the present invention, extracting the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determining whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing processing condition includes:

[0014] Calculate the distance between the target point cloud frame and the point cloud in the neighborhood grid;

[0015] Determine whether the distance is located in the search edge area, where the search edge area includes the case where the distance is equal to 0 or the boundary corresponding to the search radius;

[0016] If so, determine to perform smoothing processing between the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid

[0017] Intervals.

[0018] Optionally, in the third implementation manner of the first aspect of the present invention, extracting the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determining whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing processing condition includes: respectively determining the functions of the target point cloud frame and the point cloud frames in the neighborhood grid;

[0019] Perform a second-order derivative on the function, and determine whether the value of the second-order derivative is 0;

[0020] If so, determine to perform smoothing processing between the point cloud features of the target point cloud frame and the point cloud frame corresponding to the value of 0.

[0021] Optionally, in the fourth implementation manner of the first aspect of the present invention, the smooth transition processing of the target point cloud frame and the corresponding point cloud frames in the neighborhood grid based on the weight value to obtain an optimized point cloud map includes:

[0022] Determine the gradient values of the target point cloud frame and the corresponding point cloud frames in the neighborhood grid;

[0023] Perform coefficient assignment on the gradient values of the target point cloud frame and the neighborhood grid based on the weight value, and adjust based on the point cloud frames corresponding to the assigned gradient values to obtain an optimized point cloud map.

[0024] Optionally, in the fifth implementation manner of the first aspect of the present invention, adjusting the point cloud frame corresponding to the assigned gradient value to obtain an optimized point cloud map includes:

[0025] Performing gradient descent processing on the point cloud frame based on the assigned gradient value to obtain an optimized point cloud map.

[0026] The second aspect of the present invention provides a map optimization device, and the map optimization device includes:

[0027] An acquisition module, configured to acquire a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame, where the target point cloud frame is a point cloud frame to be optimized in the point cloud map, and the neighborhood grid includes at least one point cloud frame;

[0028] A judgment module, configured to extract point cloud features of the point cloud frames in the target point cloud frame and the neighborhood grid, and determine whether a preset smoothing processing condition is satisfied between the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid;

[0029] A calculation module, configured to determine a distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid when it is determined that the preset smoothing processing condition is satisfied, and determine a corresponding weight value based on the distance value by using a preset weight function;

[0030] An optimization module, configured to perform smooth transition processing on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map.

[0031] Optionally, in the first implementation manner of the second aspect of the present invention, the acquisition module includes:

[0032] A selection unit, configured to select at least one target point cloud frame from the point cloud map;

[0033] A search unit, configured to search for point cloud grids around each target point cloud frame as the center, and calculate the overlap rate and point cloud probability density between the searched point cloud grids;

[0034] An acquisition unit, configured to determine the neighborhood grid of the target point cloud frame based on the overlap rate and the point cloud probability density.

[0035] Optionally, in the second implementation manner of the second aspect of the present invention, the judgment module is specifically configured to:

[0036] Calculate the distance between the target point cloud frame and the point cloud in the neighborhood grid;

[0037] Judge whether the distance is located in the search edge area, where the search edge area includes the distance equal to 0 or the boundary corresponding to the search radius;

[0038] If so, determine to perform smoothing between the point cloud features of the target point cloud frame and the point cloud frames of the neighborhood grid.

[0039] Optionally, in the third implementation manner of the second aspect of the present invention, the determination module is specifically configured to:

[0040] Determine the functions of the target point cloud frame and the point cloud frames in the neighborhood grid respectively;

[0041] Perform a second-order derivative on the function, and determine whether the value of the second-order derivative is 0;

[0042] If so, determine to perform smoothing between the point cloud features of the target point cloud frame and the point cloud frame corresponding to the value of 0.

[0043] Optionally, in the fourth implementation manner of the second aspect of the present invention, the optimization module includes:

[0044] A determination unit, configured to determine the gradient values of the target point cloud frame and the corresponding point cloud frames in the neighborhood grid;

[0045] An optimization unit, configured to assign coefficients to the gradient values of the target point cloud frame and the neighborhood grid based on the weight value, and adjust based on the point cloud frames corresponding to the assigned gradient values to obtain an optimized point cloud map.

[0046] Optionally, in the fifth implementation manner of the second aspect of the present invention, the optimization unit is specifically configured to:

[0047] Perform gradient descent processing on the point cloud frame based on the assigned gradient value to obtain an optimized point cloud map.

[0048] The third aspect of the present invention provides a map optimization device, including: a memory and at least one processor, instructions are stored in the memory, and the memory and the at least one processor are interconnected by a line; the at least one processor calls the instructions in the memory so that the map optimization device executes the above-mentioned map optimization method.

[0049] The fourth aspect of the present invention provides a computer-readable storage medium, instructions are stored in the computer-readable storage medium, and when it runs on a computer, it causes the computer to execute the above-mentioned map optimization method.

[0050] In summary, the method obtains the target point cloud frame of the point cloud map and the neighborhood grid of the target point cloud frame; extracts the point cloud features of the point cloud frames in the target point cloud frame and the neighborhood grid, and determines whether the preset smoothing processing condition is satisfied between the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid; if satisfied, determines the distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determines the corresponding weight value based on the distance value by using the preset weight function; performs smoothing transition processing on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map. This application smooths two point cloud frames without feature continuity by increasing the weight value to solve the problem of discontinuous point cloud frame features in the process of making a map by the existing normal distribution change algorithm.

[0051] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention are realized and obtained by the structures specifically pointed out in the specification, the claims, and the drawings.

[0052] To make the above objectives, features, and advantages of the present invention more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, is described in detail as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of an embodiment of the map optimization method in an embodiment of the present invention;

[0054] Figure 2 It is a schematic diagram of another embodiment of the map optimization method in an embodiment of the present invention;

[0055] Figure 3 It is a schematic diagram of an embodiment of the map optimization device in an embodiment of the present invention;

[0056] Figure 4 It is a schematic diagram of another embodiment of the map optimization device in an embodiment of the present invention;

[0057] Figure 5 It is a schematic diagram of an embodiment of the map optimization device in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] To solve the above problems, this application provides a map optimization method. The method smooths the point cloud frames in the point cloud map by increasing the weight value on the basis of the point cloud map obtained by the NDT algorithm, so that the features between the two adjusted point cloud frames are continuous, avoiding the problem that the map constructed by the existing NDT algorithm has discontinuous point cloud frames, which affects subsequent positioning.

[0059] In the description, claims and the above-mentioned drawings of the present invention, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order different from that illustrated or described herein. In addition, the term "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0060] For ease of understanding, the specific process of the embodiments of the present invention will be described below. Please refer to Figure 1 , an embodiment of the map optimization method in the embodiments of the present invention. The implementation steps of this method are as follows:

[0061] 101. Obtain the target point cloud frame of the point cloud map and the neighborhood grid of the target point cloud frame.

[0062] In this step, the point cloud map is specifically generated by a lidar on the vehicle collecting road information in the area it travels through and matching the collected point clouds based on the NDT algorithm.

[0063] In this embodiment, when obtaining the point cloud map, point cloud data is collected based on the real-time position of the vehicle. Based on the collected point cloud data, as the vehicle moves, the neighborhood grid of the basic point cloud data is collected, that is, neighborhood point clouds are collected. Specifically, in the NDT algorithm, the method of adjacent grid search is used to collect point cloud data similar to the characteristics of the basic point cloud data from the moving process of the vehicle, thereby constructing the neighborhood grid. Among them, the target point cloud frame is the point cloud frame to be optimized in the point cloud map, and the neighborhood grid contains at least one point cloud frame.

[0064] In practical applications, for the acquisition of the domain grid, the neighborhood can also be calculated by overlapping grids, and the point cloud data when the vehicle moves each unit distance is obtained. Specifically, the unit distance is equal to the search distance of the lidar.

[0065] 102. Extract the point cloud features of the point cloud frames in the target point cloud frame and the neighborhood grid, and determine whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid meet the preset smoothing processing conditions.

[0066] In this embodiment, the preset smoothing processing conditions include that the distance between two point cloud frames is equal to the search radius of the neighborhood grid or zero, or the output value is zero when the second derivative of the searched point cloud frame is calculated.

[0067] Extract the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid. Specifically, calculate the actual distance between each point cloud frame searched by the neighborhood search algorithm and the target point cloud frame, and determine whether the actual distance meets the preset smoothing condition, that is, determine whether the actual distance is equal to the search radius of the neighborhood grid or zero. If so, it is determined that the currently searched point cloud frame needs to be smoothed, that is, set a transition between the searched point cloud frame and the target point cloud frame to make the point cloud frames of the overall map constructed subsequently continuous.

[0068] 103. If satisfied, determine the distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determine the corresponding weight value based on the distance value using a preset weight function.

[0069] In this embodiment, when it is determined that the searched point cloud frame meets the smoothing condition, the distance value of the point cloud frame is calculated. Of course, there are two understandings of this distance value. One is the distance between them, and the other is the cost value of the two. Among them, the cost value can be understood as the difficulty level of finding the similarity between two point cloud frames.

[0070] Furthermore, based on the distance value, calculate the weight values corresponding to the two point cloud frames through a preset weight function.

[0071] 104. Perform smooth transition processing on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map.

[0072] In this step, adjust the corresponding point cloud frame based on the calculated weight value to achieve smooth transition processing. Specifically, extract the feature gradients of the target point cloud frame and the point cloud frames in the neighborhood grid, and use the weight value to smoothly adjust the feature gradients, thereby reducing the jump level of the gradients between the two point cloud frames, or reducing parameters such as gray level and exposure degree in the two point cloud frames to achieve seamless connection between the two point cloud frames.

[0073] In summary, by setting a continuity solution of the weight value to adjust the connectivity between point cloud frames, the problem that the map constructed by the existing NDT algorithm has discontinuous point cloud frames, affecting subsequent positioning, is avoided.

[0074] Please refer to Figure 2 , for another implementation manner of the map optimization method provided by the embodiment of the present application. This method includes the following steps:

[0075] 201. Select at least one target point cloud frame from the point cloud map.

[0076] 202. Based on each target point cloud frame as the center, search the surrounding point cloud grids, and calculate the overlap rate and point cloud probability density between the searched point cloud grids.

[0077] 203. Determine the neighborhood grid of the target point cloud frame based on the overlap rate and the point cloud probability density.

[0078] In this embodiment, the selection of the target point cloud frame is specifically constructed with the point cloud located in the acquisition edge area of the lidar in the point cloud data of the current position of the vehicle as the target point cloud. Taking this target point cloud frame as the center, multiple grids are searched. Specifically, eight adjacent cells are used as a search period. From the normal distribution of the eight adjacent cells in this search period, the contribution weight of each cell is determined to determine the neighborhood grid, that is, the overlapping grid between each cell is obtained based on the contribution weight, and the overlap grid rate is calculated, as well as the point cloud density of the grid in the overlapping part. The probability of the point cloud in the overlapping part of the grid relative to the eight adjacent cells is calculated to obtain the point cloud probability density.

[0079] 204. Extract the point cloud features of the target point cloud frame and the point cloud frame in the neighborhood grid, and determine whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid meet the preset smoothing processing conditions.

[0080] In this step, when judging whether the preset smoothing processing conditions are met, it can be specifically judged by distance, or by constructing a point cloud function to find the second derivative.

[0081] In this embodiment, for the judgment based on distance, specifically, the distance between the target point cloud frame and the point cloud in the neighborhood grid is calculated; it is judged whether the distance is located in the search edge area, where the search edge area includes the distance equal to 0 or the boundary corresponding to the search radius; if so, it is determined that smoothing processing is performed between the point cloud features of the target point cloud frame and the point cloud frame of the neighborhood grid.

[0082] In practical applications, taking the search with a radius of 1m as an example, at the i-th step of optimization, there are n grids within 1m for calculating the distribution, and one of the grids (point cloud frame) exactly falls at the position of 1m. At this time, it is determined that this grid is the grid that meets the preset smoothing processing conditions. After the grid is searched at the (i + 1)-th step, the gradient value is calculated between the grid at the i-th step and the grid at the (i + 1)-th step, and the corresponding weight value is calculated. Similarly, the same thing will also happen when a grid exactly enters the search range from a position greater than 1m. Similarly, a weight is also contributed to each grid.

[0083] Furthermore, for the judgment based on the second derivative, specifically, the functions of the target point cloud frame and the point cloud frame in the neighborhood grid are respectively determined; the second derivative of the function is calculated, and it is judged whether the value of the second derivative is 0; if so, it is determined that smoothing processing is performed between the point cloud features of the target point cloud frame and the point cloud frame corresponding to the value of 0.

[0084] 205. If the condition is satisfied, determine the distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determine the corresponding weight value based on the distance value using a preset weight function.

[0085] 206. Determine the gradient values of the target point cloud frame and the corresponding point cloud frame in the neighborhood grid.

[0086] 207. Assign coefficients to the gradient values of the target point cloud frame and the neighborhood grid based on the weight value, and adjust based on the point cloud frames corresponding to the assigned gradient values to obtain an optimized point cloud map.

[0087] In this embodiment, based on the assigned gradient value, perform gradient descent processing on the point cloud frame to obtain an optimized point cloud map. Specifically, gradient descent processing using a smoothing function can be utilized.

[0088] In practical applications, this smoothing process is specifically implemented by proposing an interpolation method. Weights are calculated through distance, and a smoothing function is used to process the gradient to achieve smoothing. Among them, when calculating the weight value, it is specifically based on the weight function and the point cloud function for fusion processing to obtain a cost function. This cost function can be understood as a function for deriving point cloud similarity. Based on the cost function, a gradient matrix and a second derivative matrix can be constructed. Based on the two matrices, the corresponding fused point cloud features are calculated. Based on the fused point cloud features, the weights are determined, and then based on the weight value, the point clouds in the point cloud map are smoothed to obtain an optimized point cloud map.

[0089] In summary, by obtaining the target point cloud frame of the point cloud map and the neighborhood grid of the target point cloud frame; extracting the point cloud features of the point cloud frames in the target point cloud frame and the neighborhood grid, and determining whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing processing condition; if satisfied, determine the distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determine the corresponding weight value based on the distance value using a preset weight function; based on the weight value, perform smooth transition processing on the corresponding point cloud frames in the target point cloud frame and the neighborhood grid to obtain an optimized point cloud map. In this application, by increasing the weight value, smooth processing is performed on two point cloud frames without feature continuity to solve the problem of discontinuous point cloud frame features in the process of making a map using the existing normal distribution change algorithm.

[0090] For details, please refer to

[0091] Please refer to Figure 3 , an embodiment of the map optimization device in the embodiment of the present invention. The device includes:

[0092] An acquisition module 310, configured to acquire the target point cloud frame of the point cloud map and the neighborhood grid of the target point cloud frame, where the target point cloud frame is the point cloud frame to be optimized in the point cloud map, and the neighborhood

[0093] The grid contains at least one point cloud frame;

[0094] A judgment module 320, configured to extract the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determine whether there is

[0095] a satisfaction of a preset smoothing condition;

[0096] A calculation module 330, configured to determine a distance value between the target point

[0097] cloud frame and the corresponding point cloud frame in the neighborhood grid when it is determined that the preset smoothing condition is satisfied, and determine a corresponding weight value based on the distance value by using a preset weight function;

[0098] An optimization module 340, configured to perform a smooth transition process on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map.

[0099] In summary, by setting a continuity solution for the weight value to adjust the connectivity between point cloud frames, the problem of discontinuous point cloud frame features in the process of making a map by the existing normal distribution change algorithm is solved.

[0100] Please refer to Figure 4 , another embodiment of the map optimization device in the embodiments of the present invention, the device includes:

[0101] An acquisition module 310, configured to acquire a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame, where the target point cloud frame is a point cloud frame to be optimized in the point cloud map, and the neighborhood grid contains at least one point cloud frame;

[0102] A judgment module 320, configured to extract the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determine whether there is a satisfaction of a preset smoothing condition between the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid;

[0103] A calculation module 330, configured to determine a distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid when it is determined that the preset smoothing condition is satisfied, and determine a corresponding weight value based on the distance value by using a preset weight function;

[0104] An optimization module 340, configured to perform a smooth transition process on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map.

[0105] In this embodiment, the acquisition module 310 includes:

[0106] A selection unit 311, configured to select at least one target point cloud frame from a point cloud map;

[0107] A search unit 312, configured to search for point cloud grids around each target point cloud frame as the center, and calculate the overlap rate and point cloud probability density between the searched point cloud grids;

[0108] An acquisition unit 313, configured to determine the neighborhood grid of the target point cloud frame based on the overlap rate and the point cloud probability density.

[0109] In this embodiment, the judgment module 320 is specifically configured to:

[0110] Calculate the distance between the target point cloud frame and the point cloud in the neighborhood grid;

[0111] Judge whether the distance is located in the search edge area, where the search edge area includes the case where the distance is equal to 0 or the boundary corresponding to the search radius;

[0112] If so, determine to perform smoothing processing between the point cloud feature of the target point cloud frame and the point cloud frame in the neighborhood grid.

[0113] In this embodiment, the judgment module 320 is specifically configured to:

[0114] Respectively determine the functions of the target point cloud frame and the point cloud frame in the neighborhood grid;

[0115] Perform a second-order derivative on the function, and judge whether the value of the second-order derivative is 0;

[0116] If so, determine to perform smoothing processing between the point cloud feature of the target point cloud frame and the point cloud frame corresponding to the value of 0.

[0117] In this embodiment, the optimization module 340 includes:

[0118] A determination unit 341, configured to determine the gradient values of the target point cloud frame and the corresponding point cloud frame in the neighborhood grid;

[0119] An optimization unit 342, configured to assign coefficients to the gradient values of the target point cloud frame and the neighborhood grid based on the weight value, and perform adjustment based on the point cloud frame corresponding to the assigned gradient value to obtain an optimized point cloud map.

[0120] In this embodiment, the optimization unit 342 is specifically configured to:

[0121] Based on the assigned gradient value, perform gradient descent processing on the point cloud frame to obtain an optimized point cloud map.

[0122] In summary, the target point cloud frame of the point cloud map and the neighborhood grid of the target point cloud frame are obtained; the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid are extracted, and it is determined whether the preset smoothing processing condition is satisfied between the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid; if so, the distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid is determined, and the corresponding weight value is determined based on the distance value by using the preset weight function; the target point cloud frame and the corresponding point cloud frame in the neighborhood grid are smoothed and processed based on the weight value to obtain an optimized point cloud map. In this application, the weight value is increased to smooth the two point cloud frames without feature continuity, so as to solve the problem of discontinuous point cloud frame features in the process of making a map by the existing normal distribution change algorithm.

[0123] Please refer to Figure 5 , and a detailed description of an embodiment of the map optimization device in the embodiment of the present invention will be given from the perspective of hardware processing below.

[0124] Figure 5 FIG. is a schematic structural diagram of a map optimization device provided by an embodiment of the present invention. The map optimization device 500 may vary greatly due to configuration or performance, and may include one or more processors (central processing units, CPUs) 510 (for example, one or more processors) and a memory 520, and one or more storage media 530 for storing application programs 533 or data 532 (for example, one or more mass storage devices). Among them, the memory 520 and the storage media 530 may be transient storage or persistent storage. The program stored in the storage media 530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the map optimization device 500. Further, the processor 510 may be set to communicate with the storage media 530 and execute a series of instruction operations in the storage media 530 on the map optimization device 500.

[0125] The map optimization device 500 may further include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input / output interfaces 560, and / or one or more operating systems 531, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, and so on. Those skilled in the art can understand that Figure 5 the shown map optimization device structure does not constitute a limitation on the electronic device provided by this application, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0126] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer is caused to execute the steps of the map optimization method provided in the foregoing embodiments.

[0127] In practical applications, the method provided above can be implemented based on artificial intelligence technology. Among them, artificial intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results of theory, method, technology, and application system. Specifically, it can be executed based on a server. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery network (Content Delivery Network, CDN), and big data and artificial intelligence platforms.

[0128] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the devices and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0129] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0130] Above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for map optimization, characterized in that, The map optimization method includes the following steps: Obtain a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame, where the target point cloud frame is a point cloud frame to be optimized in the point cloud map, and the neighborhood grid contains at least one point cloud frame; Extract the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determine whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing condition; If satisfied, determine the distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determine the corresponding weight value based on the distance value using a preset weight function; Perform a smooth transition process on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map.

2. The map optimization method according to claim 1, characterized in that The step of obtaining a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame includes: Select at least one target point cloud frame from the point cloud map; Based on each target point cloud frame as the center, search for the surrounding point cloud grids, and calculate the overlap rate and point cloud probability density between the searched point cloud grids; Determine the neighborhood grid of the target point cloud frame based on the overlap rate and the point cloud probability density.

3. The map optimization method according to claim 2, wherein, The step of extracting the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determining whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing condition includes: Calculate the distance between the target point cloud frame and the point cloud in the neighborhood grid; Determine whether the distance is located in the search edge area, where the search edge area includes the case where the distance is equal to 0 or the boundary corresponding to the search radius; If so, determine to perform smoothing between the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid.

4. The map optimization method according to claim 2, wherein, The step of extracting the point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determining whether the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid satisfy a preset smoothing condition includes: Respectively determine the functions of the target point cloud frame and the point cloud frames in the neighborhood grid; Perform a second-order derivative on the function, and determine whether the value of the second-order derivative is 0; If so, determine to perform smoothing between the point cloud features of the target point cloud frame and the point cloud frame corresponding to the value of 0.

5. The map optimization method according to any one of claims 1-4, characterized in that The step of performing a smooth transition process on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map includes: Determine the gradient values of the target point cloud frame and the corresponding point cloud frame in the neighborhood grid; Assign coefficients to the gradient values of the target point cloud frame and the neighborhood grid based on the weight value, and adjust based on the point cloud frames corresponding to the assigned gradient values to obtain an optimized point cloud map.

6. The map optimization method according to claim 5, wherein The step of adjusting based on the point cloud frames corresponding to the assigned gradient values to obtain an optimized point cloud map includes: Perform gradient descent processing on the point cloud frames based on the assigned gradient values to obtain an optimized point cloud map.

7. A map optimization device, characterized in that, The map optimization device includes: An acquisition module, configured to acquire a target point cloud frame of a point cloud map and a neighborhood grid of the target point cloud frame, where the target point cloud frame is a point cloud frame to be optimized in the point cloud map, and the neighborhood grid includes at least one point cloud frame; A judgment module, configured to extract point cloud features of the target point cloud frame and the point cloud frames in the neighborhood grid, and determine whether a preset smoothing processing condition is satisfied between the point cloud features of the target point cloud frame and the point cloud features of the neighborhood grid; A calculation module, configured to, when it is determined that the preset smoothing processing condition is satisfied, determine a distance value between the target point cloud frame and the corresponding point cloud frame in the neighborhood grid, and determine a corresponding weight value based on the distance value by using a preset weight function; An optimization module, configured to perform a smoothing transition process on the target point cloud frame and the corresponding point cloud frame in the neighborhood grid based on the weight value to obtain an optimized point cloud map.

8. The map optimization device according to claim 7, wherein The optimization module includes: A determination unit, configured to determine gradient values of the target point cloud frame and the corresponding point cloud frame in the neighborhood grid; An optimization unit, configured to perform coefficient assignment on the gradient values of the target point cloud frame and the neighborhood grid based on the weight value, and adjust the point cloud frames corresponding to the assigned gradient values to obtain an optimized point cloud map.

9. A map optimization device, characterized in that, The map optimization device includes a memory and at least one processor, and instructions are stored in the memory; the at least one processor calls the instructions in the memory so that the map optimization device executes each step of the map optimization method according to any one of claims 1-6.

10. A computer-readable storage medium having instructions stored thereon, characterized in that, When the instructions are executed by the processor, each step of the map optimization method according to any one of claims 1-6 is implemented.

Images