Method for controlling vehicle, vehicle and storage medium
By acquiring the three-dimensional point cloud data of drones and vehicles, performing Gaussian conversion and parallel processing, the problem of low efficiency in the construction of three-dimensional maps in the existing technology is solved, and the rapid and accurate three-dimensional map construction is achieved, and the decision-making of assisted driving functions is improved. Efficiency and safety.
Patent Information
- Application Number
- CN202411366687.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-29
- Publication Date
- 2024-12-06
AI Technical Summary
The existing technology cannot accurately and quickly build a three-dimensional map of the road, resulting in inefficient decision-making of assisted driving functions and safety hazards, mainly due to the high correlation between single data sources and three-dimensional point cloud data, resulting in a long calculation time.
By obtaining the three-dimensional point cloud data collected by the target drone and vehicle, performing Gaussian conversion processing, multiple three-dimensional Gaussian models are obtained, and parallel processing is carried out to quickly build a three-dimensional map, reducing computing complexity and acquisition blind spots.
It realizes the accurate and rapid construction of three-dimensional maps, improves the decision-making efficiency of assisted driving functions, reduces the risk of computing complexity and collection blind spots, and improves driving safety.
Smart Images

Figure CN119078847A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of vehicle technology, and more specifically, to a method for controlling a vehicle, a vehicle, and a storage medium in the field of vehicle technology. Background Art
[0002] At present, when a vehicle is driving on the road with the assisted driving function turned on, road data is usually collected through the vehicle's own sensors; but due to the single data source, there are often blind spots in the collection, making it difficult to collect complete road data, resulting in low accuracy of the road data; and because these road data are usually three-dimensional point cloud data with a large amount of data, and the data in the three-dimensional point cloud data have a high correlation; therefore, it takes a lot of computing time to process the road data; based on this, the existing technology cannot accurately and quickly construct a three-dimensional map of the road, resulting in low decision-making efficiency of the assisted driving function and even safety hazards.
[0003] Therefore, how to accurately and quickly build three-dimensional maps to improve the decision-making efficiency of assisted driving functions is a technical problem that needs to be solved urgently. Summary of the invention
[0004] The present application provides a method for controlling a vehicle, a vehicle and a storage medium, wherein the method can accurately and quickly construct a three-dimensional map to improve the decision-making efficiency of an assisted driving function.
[0005] In a first aspect, a method for controlling a vehicle is provided, the method comprising: while the vehicle is on a target road, acquiring first three-dimensional point cloud data collected by a target unmanned aerial vehicle device and second three-dimensional point cloud data collected by the vehicle to obtain three-dimensional point cloud data of the target road; performing Gaussian transformation processing on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models; wherein the multiple three-dimensional Gaussian models correspond one-to-one to the three-dimensional point cloud data; performing parallel processing on the multiple three-dimensional Gaussian models to obtain a three-dimensional map of the target road; and controlling the vehicle to perform an assisted driving function on the target road according to the three-dimensional map.
[0006] In an embodiment of the present application, three-dimensional point cloud data is obtained by acquiring first three-dimensional point cloud data collected by a target drone and second three-dimensional point cloud data collected by a vehicle; since the drone has a larger collection range, this can reduce collection blind spots, thereby improving the integrity of the point cloud data of the road; and multiple three-dimensional Gaussian models are obtained by performing Gaussian transformation processing on each data in the three-dimensional point cloud data; since each three-dimensional Gaussian model is independent of each other, this can reduce the complexity of the calculation process; compared to the method of processing road data in the prior art, the embodiment of the present application can process multiple three-dimensional Gaussian models in parallel, thereby quickly obtaining a three-dimensional map of the target road; and on the target road, according to the three-dimensional map, control the vehicle to perform an assisted driving function; based on this, the embodiment of the present application can accurately and quickly construct a three-dimensional map to improve the decision-making efficiency of the assisted driving function.
[0007] In combination with the first aspect, in some possible implementations, multiple three-dimensional Gaussian models are processed in parallel to obtain a three-dimensional map of the target road, including: using a parallel method to map multiple three-dimensional Gaussian models to a two-dimensional plane to obtain a target two-dimensional image corresponding to each three-dimensional Gaussian model; and synthesizing multiple target two-dimensional images to obtain a three-dimensional map.
[0008] Based on the above embodiments, by processing the three-dimensional Gaussian model in a parallel manner, the speed of processing the three-dimensional Gaussian model can be improved, thereby quickly obtaining a three-dimensional map; by mapping the three-dimensional Gaussian model to a two-dimensional plane, the computational complexity of the three-dimensional Gaussian model can be reduced, thereby facilitating image processing of two-dimensional images.
[0009] In combination with the first aspect and the above-mentioned implementation methods, in some possible implementation methods, each data in the three-dimensional point cloud data includes position information and attribute information; each data has at least one adjacent data in the three-dimensional point cloud data; each data in the three-dimensional point cloud data is Gaussian transformed to obtain multiple three-dimensional Gaussian models, including: determining the center position corresponding to each data according to the position information of each data; determining the covariance matrix corresponding to each data according to the position information of each data and the position information of at least one adjacent data; determining the attribute parameters corresponding to each data according to the attribute information of each data; obtaining the three-dimensional Gaussian model of each data according to the center position corresponding to each data, the covariance matrix corresponding to each data and the attribute parameters corresponding to each data.
[0010] Based on the above embodiments, a three-dimensional Gaussian model of each data is obtained through the position information of each data in the three-dimensional point cloud data, the position information of at least one adjacent data and the attribute information of each data. This can reduce the complexity of the three-dimensional point cloud data and the calculation time, thereby achieving efficient and fast processing of the three-dimensional Gaussian model.
[0011] In combination with the first aspect and the above-mentioned implementation methods, in some possible implementation methods, Gaussian transformation processing is performed on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models, including: preprocessing the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data; wherein the preprocessing includes at least one of denoising processing and alignment processing; Gaussian transformation processing is performed on each data in the preprocessed three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models.
[0012] Based on the above embodiment, by preprocessing the three-dimensional point cloud data, noise points in the three-dimensional point cloud data can be reduced and the quality of the three-dimensional point cloud data can be improved, so as to improve the efficiency of Gaussian transformation processing on each data in the three-dimensional point cloud data.
[0013] In combination with the first aspect and the above-mentioned implementation methods, in some possible implementation methods, the method provided in the embodiment of the present application also includes: determining whether there is a first drone device with a communication connection with the vehicle; if the first drone device exists, determining the first drone device as a target drone device; if the first drone device does not exist, establishing a communication connection between the vehicle and a second drone device through a target object, and determining the second drone device as a target drone device; wherein the target object is a drone site or the first vehicle within a preset distance of the vehicle; and the second drone device has a communication connection with the target object.
[0014] Based on the above embodiment, by judging whether the vehicle has the first drone device; and in the absence of the first drone device, the vehicle establishes a communication connection with the second drone device through the target object, so as to ensure the existence of a target drone device that has a communication connection with the vehicle, so that the target drone device can be used to collect the second three-dimensional point cloud data of the target road in the future, thereby achieving the integrity of the three-dimensional point cloud data.
[0015] In combination with the first aspect and the above-mentioned implementation methods, in some possible implementation methods, a communication connection between a vehicle and a second drone device is established through a target object, including: sending a request message to the target object; wherein the request message is used to request the collection of three-dimensional point cloud data of the target road; receiving indication information of the target object; wherein the indication information carries the identification of the second drone device; and establishing a communication connection with the second drone device according to the identification of the second drone device.
[0016] Based on the above embodiment, in the absence of the first drone device, the identification of the second drone can be obtained by interacting with the target object through data, so as to establish a communication connection with the second drone device, and then the second drone device can be used as the target drone device; in this way, it can be ensured that regardless of the presence of the first drone device, the second three-dimensional point cloud data of the target road can be collected through the drone device.
[0017] In combination with the first aspect and the above-mentioned implementation methods, in some possible implementation methods, the target drone includes multiple drone devices; the method provided in the embodiment of the present application also includes: dividing the target road into multiple driving sections; determining the collection tasks of each target drone device based on the multiple driving sections; wherein each collection task includes at least one of the multiple driving sections; and sending target instructions corresponding to each target drone device to each target drone device respectively; wherein the target instructions include the collection tasks corresponding to each target drone device.
[0018] Based on the above embodiment, when there are multiple target drone devices, the target road can be divided into multiple driving sections, and the collection task of each target drone device can be determined; in this way, the collection task of the target road can be distributed to multiple target drone devices to quickly obtain the second three-dimensional point cloud data; and it can reduce errors during collection and improve the accuracy of the second three-dimensional point cloud data.
[0019] In combination with the first aspect and the above-mentioned implementation methods, in some possible implementation methods, the method provided in the embodiment of the present application also includes: receiving request information from the second vehicle through the target drone device; wherein the request information is used to request a three-dimensional map of the target road; and sending the three-dimensional map of the target road to the second vehicle through the target drone device.
[0020] Based on the above embodiment, communication between vehicles can be achieved through the target drone device; by providing the second vehicle with a three-dimensional map of the target road, the second vehicle can be prevented from processing the same second three-dimensional point cloud data again, thereby reducing the consumption of computing resources.
[0021] In a second aspect, a device for controlling a vehicle is provided, the device comprising:
[0022] An acquisition module, used to acquire the first three-dimensional point cloud data collected by the target UAV device and the second three-dimensional point cloud data collected by the vehicle when the vehicle is driving on the target road, so as to obtain the three-dimensional point cloud data of the target road;
[0023] A conversion processing module is used to perform Gaussian conversion processing on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models; wherein the multiple three-dimensional Gaussian models correspond to the three-dimensional point cloud data one by one;
[0024] A parallel processing module is used to process multiple three-dimensional Gaussian models in parallel to obtain a three-dimensional map of the target road;
[0025] The control module is used to control the vehicle to perform assisted driving functions on the target road according to the three-dimensional map.
[0026] In one possible implementation, the parallel processing module is specifically used to map multiple three-dimensional Gaussian models to a two-dimensional plane in a parallel manner to obtain a target two-dimensional image corresponding to each three-dimensional Gaussian model; and synthesize multiple target two-dimensional images to obtain a three-dimensional map.
[0027] In a possible implementation, the conversion processing module is specifically used to determine the center position corresponding to each data according to the position information of each data; determine the covariance matrix corresponding to each data according to the position information of each data and the position information of at least one adjacent data; determine the attribute parameters corresponding to each data according to the attribute information of each data; obtain the three-dimensional Gaussian model of each data according to the center position corresponding to each data, the covariance matrix corresponding to each data and the attribute parameters corresponding to each data.
[0028] In one possible implementation, the conversion processing module is specifically used to preprocess the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data; wherein the preprocessing includes at least one of denoising processing and alignment processing; Gaussian transformation processing is performed on each data in the preprocessed three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models.
[0029] In one possible implementation, the device also includes: a determination module, used to determine whether there is a first drone device in communication connection with the vehicle; if the first drone device exists, the first drone device is determined as a target drone device; if the first drone device does not exist, a communication connection is established between the vehicle and a second drone device through a target object, and the second drone device is determined as the target drone device; wherein the target object is a drone site or the first vehicle within a preset distance of the vehicle; the second drone device has a communication connection with the target object.
[0030] In a possible implementation, the device further includes: a sending module, configured to send request information to a target object; wherein the request information is used to request the collection of three-dimensional point cloud data of a target road;
[0031] In a possible implementation, the device also includes: a receiving module, used to receive indication information of the target object; wherein the indication information carries an identifier of the second drone device; and establish a communication connection with the second drone device according to the identifier of the second drone device.
[0032] In a possible implementation, the determination module is also used to divide the target road into multiple driving sections; determine the collection tasks of each target UAV device based on the multiple driving sections; wherein each collection task includes at least one of the multiple driving sections; and send target instructions corresponding to each target UAV device to each target UAV device respectively; wherein the target instructions include the collection tasks corresponding to each target UAV device.
[0033] In a possible implementation, the receiving module is further used to receive request information from the second vehicle through the target UAV device; wherein the request information is used to request a three-dimensional map of the target road;
[0034] In a possible implementation, the sending module is also used to send a three-dimensional map of the target road to the second vehicle via the target drone device.
[0035] In a third aspect, a vehicle is provided, comprising a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the vehicle executes the method in the first aspect or any possible implementation of the first aspect.
[0036] In a fourth aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation of the first aspect.
[0037] In a fifth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the above-mentioned first aspect or any possible implementation manner of the first aspect.
[0038] In a sixth aspect, a collaborative system is provided, the system comprising: a vehicle and at least one unmanned aerial vehicle device; the vehicle is respectively connected to the at least one unmanned aerial vehicle device for communication. The vehicle is used to execute the method in the first aspect or any possible implementation of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 is a schematic diagram of a vehicle and drone collaborative system provided in an embodiment of the present application;
[0040] Figure 2 is a schematic diagram of another vehicle and drone collaborative system provided in an embodiment of the present application;
[0041] Figure 3 is a schematic flow chart of a method for controlling a vehicle provided in an embodiment of the present application;
[0042] Figure 4 is a schematic flow chart of another method for controlling a vehicle provided in an embodiment of the present application;
[0043] Figure 5 is a schematic diagram of the structure of a device for controlling a vehicle provided in an embodiment of the present application;
[0044] Figure 6 It is a structural schematic diagram of a vehicle provided in an embodiment of the present application. DETAILED DESCRIPTION
[0045] The technical solution in the present application will be described clearly and in detail below in conjunction with the accompanying drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.
[0046] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.
[0047] (1) 3D Gaussian Splatting (3DGS) technology: A 3D scene construction method based on explicit representation. Its core is to use Gaussian functions to represent points or volumes in 3D space. Each point is defined by attributes such as mean, transparency, and color to form a Gaussian distribution, which determines the spatial position of the point and the degree of diffusion along each axis. This method enables 3DGS technology to capture details in the scene more accurately and generate high-quality images in real time when synthesizing multiple views.
[0048] For example, 3DGS technology can convert 3D point cloud data into an efficient and renderable 3D Gaussian model. Each 3D point data is mapped to a Gaussian body, which contains information such as position, direction, scale, opacity and color; this information is used in the subsequent rasterization and rendering process.
[0049] (2) Three-dimensional Gaussian model: In the field of three-dimensional graphics and computer vision, it is a mathematical model used to represent three-dimensional point cloud data. Specifically, the three-dimensional Gaussian model defines a voxel or particle with a specific shape, size and direction in three-dimensional space through a Gaussian distribution function, which is used to approximate the geometric and radiation information in the scene. The three-dimensional Gaussian model is defined based on the three-dimensional Gaussian distribution. Mathematically, a three-dimensional Gaussian distribution can be completely described by its mean vector (μ) and covariance matrix (Σ). The mean vector defines the center position of the Gaussian body, while the covariance matrix determines the shape and direction of the Gaussian body. In addition to geometric properties, the three-dimensional Gaussian model can also be associated with radiation properties, such as color, opacity, etc. These properties are used to simulate the propagation and scattering effects of light in three-dimensional space during the rendering process, thereby generating realistic three-dimensional images.
[0050] (3) 3D point cloud data: A spatial data set obtained by 3D scanning equipment (such as LiDAR, structured light scanner, etc.), which directly records the 3D coordinate information of the object surface or scene. Specifically, 3D point cloud data consists of a large number of points, each of which contains the coordinate position in 3D space (usually X, Y, and Z coordinate values), and sometimes may also contain other attribute information, such as color, reflectivity, intensity, etc.
[0051] At present, when a vehicle is driving on the road with the assisted driving function turned on, the road data is usually collected through the vehicle's own sensors; however, due to the single data source, there are often blind spots in the collection, making it difficult to collect complete road data, resulting in low accuracy of road data;
[0052] For example, the vehicle collects road data in real time while the vehicle is driving on the road through sensors such as cameras and LiDAR installed on the vehicle itself. However, the sensors installed on the vehicle itself are usually located at a fixed position of the vehicle (for example, in front of the vehicle); therefore, during the driving of the vehicle, it is difficult for the vehicle to perceive the situation to the side or rear of the vehicle; especially when turning or changing lanes, if there is a large vehicle blocking the lane next to it, the blocked area becomes a lateral blind spot, and the vehicle's sensors cannot effectively collect data in this area. In addition, due to the single data source, it may not be possible to fully capture every detail of the road, such as small cracks in the road surface, tiny potholes, etc. Therefore, it is difficult to collect complete road data only through the vehicle's own sensors, resulting in low accuracy of road data, which affects the judgment of the vehicle's assisted driving function on the road.
[0053] Furthermore, since these road data are usually three-dimensional point cloud data with a large amount of data and the data in the three-dimensional point cloud data are highly correlated, a lot of computing time is required when processing the road data.
[0054] For example, since road data usually contains the geometric shape of the road (e.g., curves, ramps, bridges, tunnels, etc.), material features (e.g., asphalt material, concrete material, etc.), and various abnormal data (e.g., cracks, potholes, ruts, etc. on the road), the amount of 3D point cloud data is usually large and complex, and the data has a high correlation. The traditional processing steps include: preprocessing, feature extraction, surface reconstruction, etc. of 3D point cloud data, which requires a lot of mathematical operations and spatial analysis, resulting in a huge amount of tasks and a lot of computing time.
[0055] Based on the above reasons, it can be concluded that the existing technology cannot accurately and quickly build a three-dimensional map of the road, which leads to low decision-making efficiency of the assisted driving function and even safety hazards. Therefore, how to accurately and quickly build a three-dimensional map to improve the decision-making efficiency of the assisted driving function is a technical problem that needs to be solved urgently.
[0056] In view of this, an embodiment of the present application proposes a method for controlling a vehicle, a vehicle and a storage medium; the present application obtains three-dimensional point cloud data by acquiring a target drone and a vehicle; in this way, the data source for collecting three-dimensional point cloud data can be expanded; multiple three-dimensional Gaussian models are obtained by performing Gaussian transformation processing on each data in the three-dimensional point cloud data; multiple three-dimensional Gaussian models are processed in parallel to obtain a three-dimensional map of the target road; and the vehicle is controlled to perform an assisted driving function on the target road according to the three-dimensional map; this scheme is used to accurately and quickly construct a three-dimensional map to improve the decision-making efficiency of the assisted driving function.
[0057] In order to illustrate the technical solution described in the present application, a specific embodiment is provided below. First, a vehicle and drone cooperative system according to an embodiment of the present application is introduced.
[0058] Figure 1 1 is a schematic diagram of a vehicle-drone collaboration system provided by an embodiment of the present application. The collaboration system 100 includes: a vehicle 110 and at least one drone device 120 (eg, drone device 121, drone device 122, drone device 123).
[0059] Exemplarily, the vehicle 110 has multiple types of sensors, such as cameras, laser radars, etc. These sensors can collect road data of the target road in real time when the vehicle 110 is traveling on the target road; the road data can be road surface data.
[0060] Exemplarily, the drone device 120 may be equipped with a camera, a lidar, a multispectral sensor and other equipment to collect road data of the target road.
[0061] Exemplarily, the vehicle 110 is connected to each of the plurality of drone devices 120. Through the communication connection, the vehicle 110 can send instruction information to the drone device 120 to instruct the drone device 120 to collect road data. The drone 120 can send the road data collected by the drone 120 to the vehicle 110.
[0062] Exemplarily, the communication connection may be a wireless communication connection, such as a cellular network connection, a Wi-Fi (Wireless Fidelity) connection, a Bluetooth (BlueTooth, BT) connection, etc., and the embodiments of the present application do not impose specific limitations on this.
[0063] Exemplarily, the vehicle 110 has an assisted driving function; through the assisted driving function, the vehicle 110 can control itself to travel on the target road, thereby reducing human intervention.
[0064] Optional, such as Figure 2 As shown, Figure 2 2 is a schematic diagram of another vehicle-drone cooperative system provided in an embodiment of the present application. The cooperative system 200 includes: a plurality of vehicles 210 and a plurality of drone devices 220. The plurality of vehicles 210 may include: a vehicle 211 and a vehicle 212; the plurality of drone devices 220 may include: a drone device 221, a drone device 222, a drone device 223, and a drone device 224.
[0065] Exemplarily, vehicle 210 can establish a communication connection with at least one drone device 220; drone device 220 can establish a communication connection with at least one vehicle 210. For example, vehicle 211 can establish a communication connection with drone device 221, drone device 222, and drone device 223. Vehicle 212 can establish a communication connection with drone device 223 and drone device 224.
[0066] As an example, the vehicles 210 can communicate with each other through the drone device 220; for example, since the vehicle 211 and the vehicle 212 have a communication connection with the drone device 223, the vehicle 211 and the vehicle 212 can communicate with each other through the drone device 223.
[0067] As another example, vehicles 210 may communicate with each other; a communication connection may be established between vehicle 211 and vehicle 212; and through the communication connection, mutual communication between vehicle 211 and vehicle 212 may be achieved.
[0068] For a detailed description of the vehicle 210 and the drone device 220 in the collaborative system 200, please refer to Figure 1 The vehicle 110 and the drone device 120 in the collaborative system 100 are not described in detail here.
[0069] Combine the following Figure 3 to Figure 4 The method for controlling a vehicle provided in an embodiment of the present application is described in detail.
[0070] It should be understood that the embodiments of the present application do not specifically limit the execution subject of the method for controlling a vehicle, as long as the method for controlling a vehicle according to the embodiments of the present application can be communicated by running a program that records the code of the method for controlling a vehicle according to the embodiments of the present application. For example, the execution subject of the method for controlling a vehicle provided in the embodiments of the present application may be a vehicle, or a device for controlling a vehicle applied to a vehicle, such as a chip.
[0071] Figure 3 A flowchart of a method for controlling a vehicle provided in an embodiment of the present application.
[0072] S301. When the vehicle is on the target road, obtain the first three-dimensional point cloud data collected by the target UAV device and the second three-dimensional point cloud data collected by the vehicle to obtain the three-dimensional point cloud data of the target road.
[0073] Exemplarily, the target UAV device collects the first three-dimensional point cloud data of the target road through the camera, lidar, multispectral sensor and other equipment it carries; the first three-dimensional point cloud data can be the road data in front of the vehicle. It should be explained that since the target UAV device has the ability to fly at high altitudes, it can collect the area in front of the vehicle above the vehicle; the first three-dimensional point cloud data collected by the target UAV device covers a wide range and can quickly capture the overall condition of the target road. However, due to the high flight altitude, the three-dimensional point cloud data collected by the target UAV device has a relatively low resolution in the vertical direction, and it may be difficult to capture the detailed features of the target road.
[0074] For example, the vehicle obtains 3D point cloud data of the local target road around the vehicle through sensors such as cameras and laser radars installed on the vehicle. It should be explained that since the vehicle is driving on the target road, the second 3D point cloud data collected can be details and local features of the target road; for example, abnormal data such as cracks and potholes on the target road.
[0075] It can be understood that by acquiring the first three-dimensional point cloud data collected by the target UAV device and the second three-dimensional point cloud data collected by the vehicle, the integrity of the acquired three-dimensional point cloud data can be improved.
[0076] S302. Perform Gaussian transformation on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models.
[0077] Among them, multiple three-dimensional Gaussian models correspond one-to-one to the three-dimensional point cloud data.
[0078] For example, if the three-dimensional point cloud data includes data 1 to data 5, by performing Gaussian transformation on data 1 to data 5, three-dimensional Gaussian model 1 corresponding to data 1, three-dimensional Gaussian model 2 corresponding to data 2, three-dimensional Gaussian model 3 corresponding to data 3, three-dimensional Gaussian model 4 corresponding to data 4, and three-dimensional Gaussian model 5 corresponding to data 5 can be obtained.
[0079] Exemplarily, each data in the three-dimensional point cloud data may be discrete data. The three-dimensional point cloud data may be a collection of a large amount of data in the three-dimensional space of the target road.
[0080] In actual application scenarios, three-dimensional point cloud data may include the geometric shape of the target road surface (e.g., curves, ramps, bridges, tunnels, etc.), material characteristics (e.g., asphalt material, concrete material, etc.), and various abnormal data (e.g., cracks, potholes, ruts, etc.)
[0081] Exemplarily, each data in the three-dimensional point cloud data includes position information and attribute information; each data has at least one adjacent data in the three-dimensional point cloud data.
[0082] For example, location information can be represented by a three-dimensional coordinate system, such as (x1, y2, z3). Attribute information can be used to represent the characteristics of data, such as color, reflectivity, material, etc. At least one adjacent data of data 1 in the three-dimensional point cloud data may include data 3, data 4, data 5, data 7, and data 8.
[0083] Exemplarily, Gaussian transformation processing is performed on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models, including: determining the center position corresponding to each data according to the position information of each data; determining the covariance matrix corresponding to each data according to the position information of each data and the position information of at least one adjacent data; determining the attribute parameters corresponding to each data according to the attribute information of each data; obtaining the three-dimensional Gaussian model of each data according to the center position corresponding to each data, the covariance matrix corresponding to each data and the attribute parameters corresponding to each data.
[0084] Exemplarily, the attribute parameters may include opacity parameters, color parameters, and other pre-configured parameters. The opacity parameter may be used to control the light shielding effect of the three-dimensional Gaussian model. The color parameter is used to reflect the material and lighting effect of the target road corresponding to the corresponding data point.
[0085] For example, since a three-dimensional Gaussian model is an ellipsoidal volume with a point in three-dimensional space (called the center point or mean point) as the center, a shape and direction defined by a covariance matrix, and a specific opacity and color, the three-dimensional Gaussian model can be expressed using the following formula:
[0086]
[0087] Wherein, x can represent any data in the three-dimensional point cloud data, such as data 1. μ is the center position (i.e., mean value) of the corresponding three-dimensional Gaussian model obtained by transforming the position information of the data. Σ is the covariance matrix of the corresponding three-dimensional Gaussian model obtained by transforming the position information of the data and the position information of at least one adjacent data. α can be the opacity parameter of the corresponding three-dimensional Gaussian model obtained by transforming the attribute information of the data; c can be the color of the corresponding three-dimensional Gaussian model obtained by transforming the attribute information of the data.
[0088] For example: Data 1 in the three-dimensional point cloud data is a point in a pothole area on the surface of the target road; around this point, there is also at least one adjacent data: data 3, data 4, data 5, data 7, and data 8. The position information (x1, y1, z1) of data 1 is used as the center position. The covariance matrix can be calculated by the position information (x1, y1, z1) of data 1 and the position information of data 3, data 4, data 5, data 7, and data 8; for example, the covariance matrix can be obtained by using the principal component analysis method for these data. The covariance matrix can finally obtain the shape and direction of the three-dimensional Gaussian model, which is used to reflect the actual shape of the pothole. According to the depth of the pothole and the contrast of the surrounding environment, the parameter value of the opacity parameter is determined. According to the actual material of the pothole (such as asphalt material) and possible shadow effects, the parameter value of the color parameter is determined. In summary, according to the center position, the covariance matrix, the parameter value of the opacity parameter, and the parameter value of the color parameter, the three-dimensional Gaussian model 1 corresponding to data 1 can be obtained.
[0089] Based on the above embodiments, a three-dimensional Gaussian model of each data is obtained through the position information of each data in the three-dimensional point cloud data, the position information of at least one adjacent data and the attribute information of each data. This can reduce the complexity of the three-dimensional point cloud data and the calculation time, thereby achieving efficient and fast processing of the three-dimensional Gaussian model.
[0090] Exemplarily, performing Gaussian transformation processing on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models includes: preprocessing the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data; performing Gaussian transformation processing on each data in the preprocessed three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models. The preprocessing includes at least one of denoising processing and registration processing.
[0091] For example, by performing denoising on the 3D point cloud data, noise points in the 3D point cloud data can be removed, thereby improving the quality of the 3D point cloud data. Denoising can be performed by using Gaussian filtering, bilateral filtering, and the like. For example, due to the error factors during acquisition, there are some noise points in the 3D point cloud data; removing these noise points can make the 3D point cloud data smoother without affecting the information of the 3D point cloud data.
[0092] Exemplarily, since the three-dimensional point cloud data may come from multiple perspectives for the target road, there are target data from different perspectives in the three-dimensional point cloud data; therefore, it is necessary to perform registration processing on these target data, aligning these target data to the same coordinate system to form a unified and complete three-dimensional point cloud data of the target road. For example, the target data in the three-dimensional point cloud data can be registered by an iterative nearest point algorithm. For example, there are two groups of data in the three-dimensional point cloud data, namely, data group A and data group B; data group A is the first three-dimensional point cloud data collected from the target drone device; data group B is the second three-dimensional point cloud data collected from the vehicle. The two groups of data are aligned to the same coordinate system by an iterative nearest point algorithm.
[0093] Based on the above embodiment, by preprocessing the three-dimensional point cloud data, noise points in the three-dimensional point cloud data can be reduced and the quality of the three-dimensional point cloud data can be improved, so as to improve the efficiency of Gaussian transformation processing on each data in the three-dimensional point cloud data.
[0094] S303. Process multiple three-dimensional Gaussian models in parallel to obtain a three-dimensional map of the target road.
[0095] It is understandable that, since multiple 3D Gaussian models are independent of each other, by using 3D Gaussian models to represent each data in the 3D point cloud data, each 3D Gaussian model can be calculated independently without relying on other 3D Gaussian models, and without considering the complex spatial relationship of the 3D point cloud data; therefore, different 3D Gaussian models can be processed by different processors to improve the processing speed.
[0096] Exemplarily, multiple three-dimensional Gaussian models are processed in parallel to obtain a three-dimensional map of the target road, including: using a parallel method to map the multiple three-dimensional Gaussian models to a two-dimensional plane to obtain a target two-dimensional image corresponding to each three-dimensional Gaussian model; and synthesizing the multiple target two-dimensional images to obtain a three-dimensional map.
[0097] Exemplarily, multiple three-dimensional Gaussian models are simultaneously mapped to a two-dimensional plane in parallel, so that a three-dimensional image can be converted into an image on a two-dimensional plane, thereby facilitating processing on the image on the two-dimensional plane; ultimately, a target two-dimensional image is obtained.
[0098] Exemplarily, mapping multiple three-dimensional Gaussian models to a two-dimensional plane to obtain a target two-dimensional image corresponding to each three-dimensional Gaussian model may include: mapping multiple three-dimensional Gaussian models to a two-dimensional plane to obtain an initial two-dimensional image corresponding to each three-dimensional Gaussian model; performing image processing on each initial two-dimensional image to obtain a target two-dimensional image corresponding to each initial two-dimensional image. The image processing may include: rasterization processing and rendering processing.
[0099] For example, through rasterization, the information (e.g., color parameters, opacity parameters, etc.) in the initial two-dimensional image corresponding to the three-dimensional Gaussian model can be converted into pixel-level information to obtain the rasterized initial two-dimensional image. Then, the pixel-level information in the rasterized initial two-dimensional image is rendered to improve the image quality and realism. Finally, multiple rendered initial two-dimensional images can be synthesized to obtain a three-dimensional map.
[0100] Exemplarily, a graphics processing unit (GPU) may be used to perform rasterization and rendering processing on the initial two-dimensional image.
[0101] Based on the above embodiments, by processing the three-dimensional Gaussian model in a parallel manner, the speed of processing the three-dimensional Gaussian model can be improved, thereby quickly obtaining a three-dimensional map; by mapping the three-dimensional Gaussian model to a two-dimensional plane, the computational complexity of the three-dimensional Gaussian model can be reduced, thereby facilitating image processing of two-dimensional images.
[0102] S304. Control the vehicle to perform assisted driving functions on the target road according to the three-dimensional map.
[0103] For example, while driving, the vehicle can identify road information of the target road ahead, such as geometric shape, material characteristics, and various other abnormal data, based on the three-dimensional map; since the embodiment of the present application can quickly obtain the three-dimensional map, it can adjust driving decisions (for example, deceleration, obstacle avoidance, etc.) in a timely manner according to the road information, thereby improving driving safety and enhancing the user's driving experience.
[0104] In a possible embodiment of the present application, the method provided in the embodiment of the present application also includes: determining whether there is a first drone device with a communication connection with the vehicle; if the first drone device exists, determining the first drone device as a target drone device; if the first drone device does not exist, establishing a communication connection between the vehicle and a second drone device through a target object, and determining the second drone device as a target drone device.
[0105] The target object is a drone site or a first vehicle within a preset distance of the vehicle; and the second drone device has a communication connection with the target object.
[0106] For example, the preset distance may be a value pre-configured by the vehicle, or a range set manually, or a value adjusted in real time by the vehicle according to actual conditions, and the present application embodiment does not impose specific restrictions on this. For example, the preset distance may be 1 km.
[0107] Exemplarily, a drone station can be used to manage multiple drone devices; for example, the drone station can instruct any one of the multiple drone devices to establish a communication connection with a vehicle to collect three-dimensional point cloud data of a target road.
[0108] For example, the first vehicle may be at a distance from a vehicle that is less than a preset distance; the first vehicle may have a communication connection with at least one drone device. The first vehicle may also instruct any one of the at least one drone device to establish a communication connection with the vehicle to collect three-dimensional point cloud data of the target road.
[0109] Exemplarily, the identification of the drone device may be the model of the drone device, a unique identification certificate, or may also be the physical address of the drone.
[0110] Exemplarily, the vehicle can determine whether the first drone device exists through the communication management module. For example, the communication management module can store the identifier of the first drone device. The vehicle can detect whether there is an identifier in the communication management module to determine whether there is a first drone device with a communication connection. Each identifier is used to represent a first drone device. For example, as shown in Table 1, Table 1 is a schematic table of the contents of the communication management module. Of course, Table 1 is an exemplary description for representing the communication management module; regarding the specific representation structure of the communication management module, the embodiments of the present application do not specifically limit this.
[0111] Table 1 Communication management module
[0112] sequence Logo 1 Identification A (first drone device A) 2 Identification B (first drone device B) 3 Logo C (first drone device C) ... ...
[0113] For example, the communication management module includes an identifier A, an identifier B, and an identifier C. Among them, identifier A is the identifier of the first drone device A, identifier B is the identifier of the first drone device B, and identifier C is the identifier of the first drone device C. When the vehicle detects that identifier A, identifier B, and identifier C exist in the communication management module, it can be determined that the first drone device A, the first drone device B, and the first drone device C exist; and the first drone device A, the first drone device B, and the first drone device C are determined as target drone devices.
[0114] Optionally, the vehicle may also send a detection signal to the surrounding drone devices through a specific broadcast protocol or discovery protocol (e.g., Wi-Fi detection request protocol); the detection signal is used by the vehicle to obtain the drone devices that have a communication connection with the vehicle. If a reply signal from any drone device is received, the drone device is determined to be the first drone device, and the first drone device is determined to be the target drone device.
[0115] Exemplarily, if the first drone device does not exist, establishing a communication connection between the vehicle and the second drone device through the target object includes: sending request information to the target object; receiving indication information from the target object; and establishing a communication connection with the second drone device based on an identifier of the second drone device.
[0116] The request information is used to request the collection of three-dimensional point cloud data of the target road; and the indication information carries the identification of the second UAV device.
[0117] Exemplarily, the vehicle can send a request message to the target object so that the target object knows that the vehicle needs to collect three-dimensional point cloud data of the target road through the drone device. In the case where the target object allows the vehicle to collect three-dimensional point cloud data through the drone device, a second drone device can be determined from at least one drone device that is communicatively connected to the target object. The second drone device can be a drone device that is currently in an idle state. Then, the target object sends an identifier of the second drone device to the vehicle. The target object can also forward the request message received from the vehicle to the second drone device to make it aware of the existence of the vehicle. The vehicle determines the location of the second drone device based on the identifier of the second drone device; and establishes a communication connection with the second drone device.
[0118] It can be understood that in the absence of the first drone device, the identification of the second drone can be obtained by interacting with the target object through data, so as to establish a communication connection with the second drone device, and then the second drone device can be used as the target drone device; in this way, it can be ensured that regardless of the existence of the first drone device, the second three-dimensional point cloud data of the target road can be collected through the drone device.
[0119] Based on the above embodiment, by judging whether the vehicle has the first drone device; and in the absence of the first drone device, the vehicle establishes a communication connection with the second drone device through the target object, so as to ensure the existence of a target drone device that has a communication connection with the vehicle, so that the target drone device can be used to collect the second three-dimensional point cloud data of the target road in the future, thereby achieving the integrity of the three-dimensional point cloud data.
[0120] In a possible embodiment of the present application, the target drone includes multiple drone devices; the method provided in the embodiment of the present application also includes: dividing the target road into multiple driving sections; determining the collection task of each target drone device based on the multiple driving sections; and sending target instructions corresponding to each target drone device to each target drone device respectively.
[0121] Among them, each collection task includes at least one of multiple driving sections; the target instruction includes a collection task corresponding to each target UAV device.
[0122] Exemplarily, the target instruction also includes an identifier corresponding to each target UAV device to clearly identify the target UAV device corresponding to each target instruction.
[0123] Exemplarily, the vehicle may divide the target road into multiple driving sections according to the road condition information of the target road (eg, road section type, traffic congestion conditions, road slope, etc.); different driving sections have different road condition information.
[0124] In actual scenarios, the target road can be a long highway. Due to the long mileage and complex road conditions of the target road, in order to quickly obtain the second three-dimensional point cloud data of the target road, the target road can be divided into multiple driving sections; multiple drone devices are used to collect the second three-dimensional point cloud data on some driving sections of the target road.
[0125] As an example, the collection tasks of each target drone device can be determined according to the number of multiple driving sections and the operating load of each target drone device. For example, a target drone device with a high operating load is assigned a large number of sections for collection tasks; a target drone device with a low operating load is assigned a small number of sections for collection tasks.
[0126] As another example, the collection tasks of multiple driving sections may be equally divided among the target drone devices according to the number of the multiple driving sections and the number of the target drone devices; in other words, the processing volume of the collection tasks of each target drone device is comparable.
[0127] For example: the target road includes driving section 1 to driving section 4. The section type of driving section 1 is a tunnel; there is traffic congestion on driving section 2; driving section 3 is a ring road; the number of lanes on driving section 4 is greater than the number of lanes on other driving sections. When the number of target drone devices is 3 (target drone device A, target drone device B, and target drone device C), collection tasks can be assigned to these 3 target drone devices. For example, the collection task A of target drone device A is driving section 1; the collection task B of target drone device B is: driving section 2 and driving section 3; the collection task C of target drone device C is driving section 4.
[0128] Based on the above embodiment, when there are multiple target drone devices, the target road can be divided into multiple driving sections, and the collection task of each target drone device can be determined; in this way, the collection task of the target road can be distributed to multiple target drone devices to quickly obtain the second three-dimensional point cloud data; and it can reduce errors during collection and improve the accuracy of the second three-dimensional point cloud data.
[0129] In a possible embodiment of the present application, the method provided in the embodiment of the present application also includes: receiving request information from the second vehicle through the target drone device; and sending a three-dimensional map of the target road to the second vehicle through the target drone device.
[0130] The request information is used to request a three-dimensional map of a target road.
[0131] Exemplarily, the request information may include information field 1 for requesting a three-dimensional map of a target road from a vehicle and field 2 for the identification of the second vehicle. Alternatively, the request information may also include information field 1 for the three-dimensional map of the target road and field 2 for the identification of the second vehicle; when the vehicle obtains the request information, it can be learned that the second vehicle requests a three-dimensional map of the target road by reading the content of the request information.
[0132] Exemplarily, the second vehicle may be a vehicle traveling on the target road. For example, the second vehicle may travel behind the vehicle in the embodiment of the present application on the target road. In other words, the second vehicle also needs to obtain a three-dimensional map of the target road.
[0133] For example, since each drone device can establish a communication connection with at least one vehicle, data interaction can be achieved between at least one vehicle (e.g., a vehicle and a second vehicle) that is in communication connection with the target drone device. In this way, the second vehicle can send a request message to the target drone device, which is forwarded to the vehicle by the target drone device. When the vehicle receives the request message, it can send a three-dimensional map of the target road to the second vehicle, so as to avoid processing the same second three-dimensional point cloud data again.
[0134] It can be understood that for the target drone device, it is possible to collect the second three-dimensional point cloud data of the target road for at least one vehicle on the target road, which can reduce unnecessary collection times, improve collection efficiency, and reduce resource loss of the drone device.
[0135] Based on the above embodiment, communication between vehicles can be achieved through the target drone device; by providing the three-dimensional map of the target road to the second vehicle, the second vehicle can be prevented from processing the same second three-dimensional point cloud data again, thereby reducing the consumption of computing resources.
[0136] In an embodiment of the present application, three-dimensional point cloud data is obtained by acquiring first three-dimensional point cloud data collected by a target drone and second three-dimensional point cloud data collected by a vehicle; since the drone has a larger collection range, this can reduce collection blind spots, thereby improving the integrity of the point cloud data of the road; and multiple three-dimensional Gaussian models are obtained by performing Gaussian transformation processing on each data in the three-dimensional point cloud data; since each three-dimensional Gaussian model is independent of each other, this can reduce the complexity of the calculation process; compared to the method of processing road data in the prior art, the embodiment of the present application can process multiple three-dimensional Gaussian models in parallel, thereby quickly obtaining a three-dimensional map of the target road; and on the target road, according to the three-dimensional map, control the vehicle to perform an assisted driving function; based on this, the embodiment of the present application can accurately and quickly construct a three-dimensional map to improve the decision-making efficiency of the assisted driving function.
[0137] It can be understood that the 3DGS technology is used in the embodiment of the present application to convert the collected three-dimensional point cloud data into a three-dimensional Gaussian model; and the three-dimensional Gaussian model is processed to finally obtain a three-dimensional map. Since the 3DGS technology adopts explicit representation and highly parallel workflow, compared with the prior art method of directly extracting features from three-dimensional point cloud data; and using complex query operations in implicit representation methods to process three-dimensional point cloud data to obtain a three-dimensional map, the embodiment of the present application can significantly reduce the amount of calculation; for example, when processing a large amount of three-dimensional point cloud data, each data is converted into a three-dimensional Gaussian model, and these three-dimensional Gaussian models can be processed in parallel, greatly speeding up the calculation speed. In addition, by converting the three-dimensional Gaussian model into a two-dimensional image, the effective area in the two-dimensional image can be processed, avoiding invalid calculations on blank space, thereby optimizing rendering performance.
[0138] Figure 4 It is a schematic flowchart of another method for controlling a vehicle provided in an embodiment of the present application.
[0139] For example, Figure 4 The method for controlling a vehicle shown may be executed by the vehicle, or may be executed by a device for controlling the vehicle in the vehicle, such as a chip.
[0140] like Figure 4 As shown, the vehicle control method includes S401 to S414, and S401 to S414 are described in detail below.
[0141] S401. Turn on the assisted driving function.
[0142] Exemplarily, the assisted driving function is used to control the vehicle to travel on a target road.
[0143] S402. Determine whether there is a first drone device that is in communication connection with the vehicle. If yes, execute S403; if no, execute S404.
[0144] Exemplarily, it is determined whether there is a first drone device in communication connection with the vehicle; if there is the first drone device, the first drone device can be controlled to collect three-dimensional point cloud data on the target road; if there is no first drone device, a request can be made from the drone site to establish a communication connection with the second drone device.
[0145] Exemplarily, the communication connection may be a wireless communication connection; for example, a cellular network connection, a Wi-Fi connection, a Bluetooth connection, etc.
[0146] S403. Send a control instruction to the first drone device.
[0147] Exemplarily, the control instruction is used to instruct the first drone device to collect first three-dimensional point cloud data on the target road.
[0148] Exemplarily, the first drone device can collect the first three-dimensional point cloud data on the target road by using devices such as a camera, a laser radar, and a multispectral sensor loaded on the first drone device.
[0149] S404. Request the drone site to obtain the second drone device.
[0150] Exemplarily, the drone site may be a site whose distance from the vehicle is less than a preset distance. The drone site may be used to manage at least one second drone device.
[0151] Optionally, the vehicle may also request a first vehicle, which is at a distance less than a preset distance from the vehicle, to obtain a second drone device.
[0152] S405. Establish a communication connection with the second drone device.
[0153] Exemplarily, by establishing a communication connection with the second drone device, data interaction between the vehicle and the second drone device can be achieved.
[0154] S406. Send a control instruction to the second drone device.
[0155] Exemplarily, the control instruction is used to instruct the second UAV device to collect first three-dimensional point cloud data on the target road.
[0156] Exemplarily, the second drone device can collect the first three-dimensional point cloud data on the target road by using a camera, a laser radar, a multispectral sensor, or other equipment loaded on the second drone device.
[0157] S407. While the vehicle is traveling on the target road, collect second three-dimensional point cloud data.
[0158] For example, the vehicle can obtain three-dimensional point cloud data of a local target road around the vehicle through sensors such as cameras and lidars installed on the vehicle.
[0159] S408. Integrate the first three-dimensional point cloud data and the second three-dimensional point cloud data to obtain three-dimensional point cloud data.
[0160] S409. Perform denoising and registration processing on the three-dimensional point cloud data.
[0161] Exemplarily, by performing denoising processing on the three-dimensional point cloud data, noise points in the three-dimensional point cloud data can be removed, thereby improving the quality of the three-dimensional point cloud data.
[0162] Exemplarily, by performing registration processing on these three-dimensional point cloud data, the three-dimensional point cloud data can be aligned to the same three-dimensional coordinate system.
[0163] S410. Perform Gaussian transformation on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models.
[0164] For example,
[0165] S411. Use a parallel approach to map multiple three-dimensional Gaussian models to a two-dimensional plane to obtain multiple initial two-dimensional images.
[0166] S412. Perform image processing on the multiple initial two-dimensional images to obtain multiple target two-dimensional images.
[0167] Exemplarily, image processing includes rasterization processing and rendering processing. Through rasterization processing, information (e.g., color parameters, opacity parameters, etc.) in the initial two-dimensional image corresponding to the three-dimensional Gaussian model can be converted into pixel-level information; through rendering processing, the quality of the initial two-dimensional image can be improved to obtain a target two-dimensional image.
[0168] S413. Synthesize multiple target two-dimensional images to obtain a three-dimensional map.
[0169] S414. Control the vehicle to perform assisted driving functions on the target road according to the three-dimensional map.
[0170] For example, while driving, the vehicle can identify road information of the target road ahead, such as geometric shape, material characteristics, and various other abnormal data, based on the three-dimensional map; since the embodiment of the present application can quickly obtain the three-dimensional map, it can adjust driving decisions (for example, deceleration, obstacle avoidance, etc.) in a timely manner according to the road information, thereby improving driving safety and enhancing the user's driving experience.
[0171] For detailed description of S401 to S414, please refer to Figure 3 The relevant descriptions will not be repeated here.
[0172] In an embodiment of the present application, it is determined whether there is a first drone device; and when the first drone device does not exist, a request is made to the drone site to establish a communication connection with the second drone device; this ensures that there is a drone device that can collect first three-dimensional point cloud data on the target road; this can expand the data source for collecting three-dimensional point cloud data; and the drone has a larger collection range, which can reduce collection blind spots and thus improve the integrity of road data; then, by performing Gaussian transformation processing on each data in the three-dimensional point cloud data collected by the drone device and the vehicle, multiple three-dimensional Gaussian models are obtained; since each three-dimensional Gaussian model is independent of each other, this can reduce the complexity of the calculation process; compared with the method of processing road data in the prior art, the embodiment of the present application can process multiple three-dimensional Gaussian models in parallel, so as to quickly obtain a three-dimensional map of the target road; and control the vehicle to perform the assisted driving function on the target road according to the three-dimensional map; based on this, the embodiment of the present application can accurately and quickly construct a three-dimensional map to improve the decision-making efficiency of the assisted driving function.
[0173] Combination of the above Figure 3 to Figure 4 The method for controlling a vehicle provided by the embodiment of the present application is described in detail; Figure 5 and Figure 6 The device embodiments of the present application are described in detail. It should be understood that the device in the embodiments of the present application can execute the various methods of the aforementioned embodiments of the present application, that is, the specific working processes of the following various products can refer to the corresponding processes in the aforementioned method embodiments.
[0174] Combine the following Figure 5 The device for controlling a vehicle provided in an embodiment of the present application is described in detail.
[0175] like Figure 5 As shown, Figure 5 It is a structural schematic diagram of a device for controlling a vehicle provided in an embodiment of the present application.
[0176] For example, Figure 5 As shown, the device comprises:
[0177] The acquisition module 510 is used to acquire the first three-dimensional point cloud data collected by the target UAV device and the second three-dimensional point cloud data collected by the vehicle when the vehicle is driving on the target road, so as to obtain the three-dimensional point cloud data of the target road;
[0178] The conversion processing module 520 is used to perform Gaussian conversion processing on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models; wherein the multiple three-dimensional Gaussian models correspond to the three-dimensional point cloud data one by one;
[0179] A parallel processing module 530 is used to perform parallel processing on multiple three-dimensional Gaussian models to obtain a three-dimensional map of the target road;
[0180] The control module 540 is used to control the vehicle to perform the assisted driving function on the target road according to the three-dimensional map.
[0181] In one possible implementation, the parallel processing module 530 is specifically used to map multiple three-dimensional Gaussian models to a two-dimensional plane in a parallel manner to obtain a target two-dimensional image corresponding to each three-dimensional Gaussian model; and synthesize multiple target two-dimensional images to obtain a three-dimensional map.
[0182] In a possible implementation, the conversion processing module 520 is specifically used to determine the center position corresponding to each data according to the position information of each data; determine the covariance matrix corresponding to each data according to the position information of each data and the position information of at least one adjacent data; determine the attribute parameters corresponding to each data according to the attribute information of each data; and obtain a three-dimensional Gaussian model of each data according to the center position corresponding to each data, the covariance matrix corresponding to each data, and the attribute parameters corresponding to each data.
[0183] In one possible implementation, the conversion processing module 520 is specifically used to preprocess the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data; wherein the preprocessing includes at least one of denoising processing and alignment processing; Gaussian transformation processing is performed on each data in the preprocessed three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models.
[0184] In one possible implementation, the device also includes: a determination module, used to determine whether there is a first drone device in communication connection with the vehicle; if the first drone device exists, the first drone device is determined as a target drone device; if the first drone device does not exist, a communication connection is established between the vehicle and a second drone device through a target object, and the second drone device is determined as the target drone device; wherein the target object is a drone site or the first vehicle within a preset distance of the vehicle; the second drone device has a communication connection with the target object.
[0185] In a possible implementation, the device further includes: a sending module, configured to send request information to a target object; wherein the request information is used to request the collection of three-dimensional point cloud data of a target road;
[0186] In a possible implementation, the device also includes: a receiving module, used to receive indication information of the target object; wherein the indication information carries an identifier of the second drone device; and establish a communication connection with the second drone device according to the identifier of the second drone device.
[0187] In a possible implementation, the determination module is also used to divide the target road into multiple driving sections; determine the collection tasks of each target UAV device based on the multiple driving sections; wherein each collection task includes at least one of the multiple driving sections; and send target instructions corresponding to each target UAV device to each target UAV device respectively; wherein the target instructions include the collection tasks corresponding to each target UAV device.
[0188] In a possible implementation, the receiving module is further used to receive request information from the second vehicle through the target UAV device; wherein the request information is used to request a three-dimensional map of the target road;
[0189] In a possible implementation, the sending module is also used to send a three-dimensional map of the target road to the second vehicle via the target drone device.
[0190] It should be noted that the device for controlling a vehicle provided in the above-mentioned embodiment only uses the division of the above-mentioned functional modules as an example when executing the method for controlling a vehicle. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.
[0191] In addition, the device for controlling a vehicle and the method for controlling a vehicle provided in the above embodiments belong to the same concept. Therefore, for details not disclosed in the device embodiments of this specification, please refer to the method for controlling a vehicle embodiments described above in this specification, and no further details will be given here.
[0192] Figure 6 It is a structural schematic diagram of a vehicle provided in an embodiment of the present application.
[0193] For example, Figure 6 As shown, the vehicle 600 includes: a memory 601 and a processor 602, wherein the memory 601 stores an executable program code 6011, and the processor 602 is used to call and execute the executable program code 6011 to perform a method for controlling a vehicle.
[0194] In addition, an embodiment of the present application also protects a device, which may include a memory and a processor, wherein the memory stores an executable program code, and the processor is used to call and execute the executable program code to execute a method for controlling a vehicle provided in an embodiment of the present application.
[0195] In this embodiment, the functional modules of the device can be divided according to the above method example. For example, each functional module can be corresponded, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation.
[0196] In the case of dividing each functional module according to each function, the device may also include an acquisition module, a conversion processing module, a parallel processing module, a control module, etc. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, which will not be repeated here.
[0197] It should be understood that the device provided in this embodiment is used to execute the above-mentioned method of controlling a vehicle, and thus can achieve the same effect as the above-mentioned implementation method.
[0198] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to a vehicle, the processing module may be used to control and manage the actions of the vehicle. The storage module may be used to support the vehicle in executing related program codes, etc.
[0199] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits shown in conjunction with the disclosure of the present application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.
[0200] In addition, the device provided in the embodiments of the present application may specifically be a chip, a component or a module, and the chip may include a connected processor and a memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute a method for controlling a vehicle provided in the above embodiments.
[0201] This embodiment also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is run on a computer, the computer executes the above-mentioned related method steps to implement a method for controlling a vehicle provided in the above embodiment.
[0202] This embodiment also provides a computer program product. When the computer program product is run on a computer, the computer is caused to execute the above-mentioned related steps to implement a method for controlling a vehicle provided in the above embodiment.
[0203] Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment is used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here.
[0204] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.
[0205] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0206] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which 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.
Claims
1. A method for controlling a vehicle, characterized in that: The method comprises: When the vehicle is traveling on the target road, first three-dimensional point cloud data collected by the target UAV device and second three-dimensional point cloud data collected by the vehicle are acquired to obtain the three-dimensional point cloud data of the target road; Performing Gaussian transformation processing on each data in the three-dimensional point cloud data to obtain a plurality of three-dimensional Gaussian models; wherein the plurality of three-dimensional Gaussian models correspond one-to-one to the three-dimensional point cloud data; Processing the multiple three-dimensional Gaussian models in parallel to obtain a three-dimensional map of the target road; The vehicle is controlled to perform an assisted driving function on the target road according to the three-dimensional map.
2. The method according to claim 1, characterized in that The step of performing parallel processing on the multiple three-dimensional Gaussian models to obtain the three-dimensional map of the target road includes: In a parallel manner, the multiple three-dimensional Gaussian models are mapped to a two-dimensional plane to obtain a target two-dimensional image corresponding to each three-dimensional Gaussian model; The multiple target two-dimensional images are synthesized to obtain the three-dimensional map.
3. The method according to claim 1, characterized in that Each data in the three-dimensional point cloud data includes position information and attribute information; each data has at least one adjacent data in the three-dimensional point cloud data; The Gaussian transformation is performed on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models, including: Determine the center position corresponding to each data according to the position information of each data; Determine a covariance matrix corresponding to each of the data according to the position information of each of the data and the position information of the at least one adjacent data; Determine attribute parameters corresponding to each data according to the attribute information of each data; A three-dimensional Gaussian model of each data is obtained according to the central position corresponding to each data, the covariance matrix corresponding to each data and the attribute parameter corresponding to each data.
4. The method according to claim 1, characterized in that: The Gaussian transformation is performed on each data in the three-dimensional point cloud data to obtain multiple three-dimensional Gaussian models, including: Preprocessing the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data; wherein the preprocessing includes at least one of denoising processing and registration processing; The Gaussian transformation process is performed on each data in the preprocessed three-dimensional point cloud data to obtain the multiple three-dimensional Gaussian models.
5. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Determining whether the vehicle has a first drone device in communication connection; If the first UAV device exists, determining the first UAV device as the target UAV device; If the first drone device does not exist, a communication connection is established between the vehicle and the second drone device through a target object, and the second drone device is determined as the target drone device; wherein the target object is a drone site or the first vehicle within a preset distance of the vehicle; and the second drone device has a communication connection with the target object.
6. The method according to claim 5, characterized in that The establishing of a communication connection between the vehicle and the second UAV device through the target object includes: Sending a request message to the target object; wherein the request message is used to request the collection of three-dimensional point cloud data of the target road; Receiving indication information of the target object; wherein the indication information carries an identification of the second UAV device; The communication connection is established with the second drone device according to the identifier of the second drone device.
7. The method according to any one of claims 1 to 4, characterized in that The target drone includes a plurality of drone devices; the method further includes: Dividing the target road into a plurality of driving sections; Determine, according to the multiple driving sections, a collection task for each of the target UAV devices; wherein each of the collection tasks includes at least one of the multiple driving sections; Send target instructions corresponding to each target UAV device to each target UAV device respectively; wherein the target instructions include collection tasks corresponding to each target UAV device.
8. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Receiving request information from a second vehicle through the target UAV device; wherein the request information is used to request a three-dimensional map of the target road; The three-dimensional map of the target road is sent to the second vehicle through the target UAV device.
9. A vehicle, characterized in that: The vehicle comprises: A memory for storing executable program codes; A processor, configured to call and run the executable program code from the memory, so that the vehicle executes the method according to any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.
