Vehicle posture optimization and correction method, device, server, vehicle and medium

Abstract

The embodiment of the present application provides a vehicle posture optimization and correction method, device, server, vehicle and medium, which relates to the field of vehicle positioning technology. By obtaining the self-posture, real-time dynamic signal, and relative posture uploaded by multiple vehicles, the relative posture is the relative posture between each vehicle in the multiple vehicles and the adjacent vehicles of each vehicle, and the multiple vehicles include the first type of vehicles with weak real-time dynamic signals and the second type of vehicles with good real-time dynamic signals and adjacent to the first type of vehicles; the self-posture of the first type of vehicles, the self-posture of the second type of vehicles, the relative posture between the first type of vehicles and the second type of vehicles uploaded by the first type of vehicles and the second type of vehicles, and the real-time dynamic signal of each vehicle in the first type of vehicles and the second type of vehicles are added to the factor graph; the factor graph is optimized to obtain the optimized posture of each vehicle in the first type of vehicles and the second type of vehicles, which can take into account the positioning accuracy and cost of vehicles in the weak real-time dynamic signal scenario.

Description

Technical Field

[0001] Embodiments of the present application relate to the field of vehicle positioning technology, and in particular, to a vehicle posture optimization and correction method, device, server, vehicle, and medium. Background Art

[0002] Currently, image sensors, wheel speed sensors, inertial measurement units (IMUs) and global positioning systems (GPSs) are commonly used for multi-sensor fusion to perform vehicle positioning.

[0003] In weak real-time kinematic (RTK) signal scenarios, such as those in urban buildings, underground tunnels, and overpasses, vehicle positioning information is particularly critical for driving. However, in weak RTK signal scenarios, the poor positioning accuracy of the global positioning system leads to poor RTK signals. In this case, the multi-sensor fusion method can only fuse data obtained from the inertial measurement unit, image sensor, and wheel speed sensor. This will cause the vehicle positioning accuracy to gradually deteriorate over time, or even become impossible to locate. In other words, in weak RTK signal scenarios, the accuracy of vehicle positioning using multi-sensor fusion is poor.

[0004] In order to solve the problem of poor vehicle positioning accuracy in weak real-time dynamic signal scenarios, advanced unmanned driving solutions use a method of matching laser point clouds with high-precision map point clouds to perform vehicle positioning. Although this method improves vehicle positioning accuracy in weak real-time dynamic signal scenarios, high-precision maps and multi-line laser radars with point cloud information are expensive, making it difficult to mass-produce vehicles.

[0005] In summary, in weak real-time dynamic signal scenarios, the current vehicle positioning solutions cannot balance vehicle positioning accuracy and cost. Summary of the Invention

[0006] The embodiments of the present application provide a vehicle posture optimization and correction method, device, server, vehicle and medium to improve the above-mentioned problems.

[0007] In the first aspect, an embodiment of the present application provides a vehicle posture optimization method. The method includes: obtaining the own posture, real-time dynamic signal and relative posture uploaded by multiple vehicles, wherein the relative posture is the relative posture between the vehicle uploading the relative posture and the adjacent vehicle of the vehicle, and the multiple vehicles include first-class vehicles and second-class vehicles, the first-class vehicles are in a weak real-time dynamic signal state, and the second-class vehicles are in a good real-time dynamic signal state and are adjacent to the first-class vehicles; the own posture of the first-class vehicle and the own posture of the second-class vehicle are used as variables of the factor graph, and the relative postures uploaded by the first-class vehicle and the second-class vehicle, the real-time dynamic signal of the first-class vehicle and the real-time dynamic signal of the second-class vehicle are used as factors of the factor graph to form a factor graph, wherein the factor graph is an estimation model for solving the probability of the variable given the factors; the factor graph is optimized to obtain the optimized posture of each vehicle in the first-class vehicle and the second-class vehicle.

[0008] In a second aspect, embodiments of the present application provide a vehicle posture correction method. The method comprises: obtaining and uploading the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its neighboring vehicles to a server, so that the server optimizes the vehicle posture according to the vehicle posture optimization method provided in the first aspect of the present application; when the vehicle is in a weak real-time dynamic signal state, obtaining the optimized vehicle posture sent by the server; and correcting the vehicle posture based on the optimized vehicle posture.

[0009] In a third aspect, an embodiment of the present application provides a vehicle posture optimization device. The device includes: a data acquisition module for acquiring the self-posture, real-time dynamic signal and relative posture uploaded by multiple vehicles, wherein the relative posture is the relative posture between the vehicle uploading the relative posture and the adjacent vehicle of the vehicle, and the multiple vehicles include first-class vehicles and second-class vehicles, the first-class vehicles are in a weak real-time dynamic signal state, and the second-class vehicles are in a good real-time dynamic signal state and are adjacent to the first-class vehicles; a data processing module for using the self-posture of the first-class vehicles and the self-posture of the second-class vehicles as variables of a factor graph, and using the relative postures uploaded by the first-class vehicles and the second-class vehicles, the real-time dynamic signals of the first-class vehicles and the real-time dynamic signals of the second-class vehicles as factors of the factor graph to form a factor graph, wherein the factor graph is an estimation model for solving the probability of the variables given the factors; a posture optimization module for optimizing the factor graph to obtain the optimized posture of each vehicle in the first-class vehicles and the second-class vehicles.

[0010] In a fourth aspect, an embodiment of the present application provides a vehicle posture correction device. The device includes: a data upload module for acquiring and uploading the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its neighboring vehicles to a server, so that the server optimizes the vehicle posture according to the vehicle posture optimization method provided in the first aspect of the embodiment of the present application; a data acquisition module for acquiring the optimized vehicle posture sent by the server when the vehicle is in a weak real-time dynamic signal state; and a posture correction module for correcting the vehicle posture based on the optimized vehicle posture.

[0011] In a fifth aspect, an embodiment of the present application provides a server. The server includes a memory, one or more processors, and one or more applications. The one or more applications are stored in the memory and are configured to, when called by the one or more processors, cause the server to execute the vehicle posture optimization method provided in the first aspect of the embodiment of the present application.

[0012] In a sixth aspect, an embodiment of the present application provides a vehicle. The vehicle includes a memory, one or more processors, and one or more applications. The one or more applications are stored in the memory and, when invoked by the one or more processors, cause the vehicle to execute the vehicle posture correction method provided in the second aspect of the embodiment of the present application.

[0013] In a seventh aspect, embodiments of the present application provide a computer-readable storage medium having program code stored thereon, the program code being configured to, when invoked by the one or more processors, cause a server to execute the vehicle posture optimization method provided in the first aspect of the embodiments of the present application, or cause a vehicle to execute the vehicle posture correction method provided in the second aspect of the embodiments of the present application.

[0014] The vehicle posture optimization and correction method, device, server, vehicle and medium provided in the embodiment of the present application, by taking the self-posture of the first type of vehicle and the self-posture of the second type of vehicle as the variables of the factor graph, taking the relative posture uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signal of the first type of vehicle and the real-time dynamic signal of the second type of vehicle as the factors of the factor graph, forming a factor graph, and optimizing the posture of each vehicle in the first type of vehicle and the second type of vehicle, thereby providing an optimized posture for the first type of vehicle with weak real-time dynamic signal, so that the first type of vehicle adopts the optimized self-posture to perform posture correction, thereby improving the positioning accuracy of the vehicle in the weak real-time dynamic signal scenario, and improving the accuracy of vehicle navigation. In addition, the vehicle posture optimization method can be implemented by the equipment installed on the vehicle, without the need to add expensive sensing equipment (for example, the above-mentioned multi-line laser radar). In other words, the vehicle posture optimization method can take into account the positioning accuracy and cost of the vehicle in the weak real-time dynamic signal scenario at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 This is a structural diagram of a vehicle posture correction system provided by an embodiment of the present application;

[0017] Figure 2 1 is a flow chart of a vehicle posture optimization method provided in one embodiment of the present application;

[0018] Figure 3 is a schematic diagram of a road and a vehicle provided by an exemplary embodiment of the present application;

[0019] Figure 4 is a schematic diagram of a factor graph provided by an exemplary embodiment of the present application;

[0020] Figure 5 1 is a flow chart of a vehicle posture correction method provided in one embodiment of the present application;

[0021] Figure 6 1 is a flow chart of a vehicle posture correction method provided by an embodiment of the present application;

[0022] Figure 7 1 is a flow chart of a vehicle posture correction method provided in one embodiment of the present application;

[0023] Figure 8 1 is a schematic structural diagram of a vehicle posture optimization device provided in one embodiment of the present application;

[0024] Figure 9 1 is a schematic structural diagram of a vehicle posture correction device provided in one embodiment of the present application;

[0025] Figure 10 This is a schematic diagram of the structure of a server provided in one embodiment of the present application;

[0026] Figure 11 is a structural schematic diagram of a vehicle provided in one embodiment of the present application;

[0027] Figure 12 It is a structural diagram of a computer-readable storage medium provided in one embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0029] Figure 1 FIG1 is a schematic diagram of the structure of a vehicle posture correction system provided by an exemplary embodiment of the present application. Vehicle posture correction system 100 may include a vehicle 110 and a server 120. Vehicle 110 and server 120 can communicate with each other, with vehicle 110 uploading data to server 120 and server 120 also sending data to vehicle 110.

[0030] Vehicle 110 may be, but is not limited to, a gasoline vehicle or an electric vehicle, wherein an electric vehicle may be, but is not limited to, a pure electric vehicle, a hybrid vehicle, or a fuel cell vehicle. Vehicle 110 may include one or more sensing devices. The sensing device may be used to sense information around the vehicle, such as other vehicles and obstacles near the vehicle. The sensing device may also be used to measure the distance to other objects, such as the distance between the vehicle and an obstacle. The sensing device may include, but is not limited to, an image sensor (e.g., a camera), a wheel speed sensor, an ultrasonic sensor, a millimeter wave sensor, and a laser sensor.

[0031] The vehicle 110 may also include an inertial navigation module, which may be used to locate the vehicle 110. The inertial navigation module may include one or more inertial navigation systems and one or more global navigation satellite systems (GNSS). The inertial navigation system may include an inertial measurement unit, which may include but is not limited to an accelerometer and a gyroscope. The global satellite navigation system may receive satellite signals and, at the same time, receive real-time dynamic signals sent by a reference station through a wireless receiving device. Then, based on the principle of relative positioning, it may calculate the three-dimensional coordinates of the vehicle and their accuracy in real time. Global satellite navigation systems may include, but are not limited to, the Global Positioning System (GPS), the Global Navigation Satellite System (Glonass), the Galileo Satellite Navigation System (Galileo), the Beidou Satellite Navigation System, the Wide Area Augmentation System (WAAS), the European Geostationary Navigation Overlay Service (EGNOS), and the Multi-Functional Satellite Augmentation System (MSAS).

[0032] The server 120 may be a backend server of the vehicle 110. The server 120 may be, but is not limited to, a cloud server or a traditional server. The server 120 may be, but is not limited to, a database server, a file server, a web server, a server supporting the File Transfer Protocol (FTP), a domain server, and the like.

[0033] Figure 2 1 is a flow chart of a vehicle posture optimization method provided by an embodiment of the present application. The vehicle posture optimization method can be applied to a server or a vehicle posture optimization device. The vehicle posture optimization method can include the following steps S110 to S130.

[0034] Step S110, obtaining the self-position, real-time dynamic signal and relative position uploaded by multiple vehicles, wherein the relative position is the relative position between the vehicle uploading the relative position and the adjacent vehicles of the vehicle. The multiple vehicles include first-class vehicles and second-class vehicles. The first-class vehicles are in a weak real-time dynamic signal state, and the second-class vehicles are in a good real-time dynamic signal state and are adjacent to the first-class vehicles.

[0035] The self-position in the embodiment of the present application may refer to the position and posture of the vehicle itself, and the relative position and posture may refer to the relative position and posture between any vehicle and its adjacent vehicles.

[0036] The weak real-time dynamic signal state in the embodiments of the present application can refer to a state in which the real-time dynamic signal of the vehicle is poor, and can also be understood as a state in which the accuracy of the real-time dynamic signal solution is poor, wherein the accuracy of the real-time dynamic signal solution can refer to the distance between the vehicle coordinates calculated based on the real-time dynamic signal and the actual vehicle coordinates. The smaller the distance, the higher the accuracy of the real-time dynamic signal solution. For example, the weak real-time dynamic signal state can refer to a state in which the distance between the vehicle coordinates calculated based on the real-time dynamic signal and the actual vehicle coordinates is greater than a preset distance, wherein the preset distance can be 2 cm or 3 cm.

[0037] In the embodiments of the present application, a good real-time dynamic signal state may refer to a state in which the real-time dynamic signal of the vehicle is in good condition, or may be understood as a state in which the accuracy of the real-time dynamic signal solution is good. For example, a good real-time dynamic signal state may refer to a state in which the distance between the vehicle coordinates calculated based on the real-time dynamic signal and the actual vehicle coordinates is less than or equal to the predetermined distance.

[0038] The vehicles in the first and second categories of vehicles may refer to all vehicles in the first and second categories of vehicles. Figure 3 The examples shown are for explaining the first type of vehicle, the second type of vehicle, and the vehicles in the first type of vehicle and the second type of vehicle. Figure 3As shown, assuming a tunnel is the weak real-time dynamic signal scenario, vehicles 3-6 are located within the tunnel. Since vehicles in tunnels typically experience weak real-time dynamic signals, vehicles 3-6 can be considered to be in a weak real-time dynamic signal state, meaning they belong to the first category. Vehicles 1, 2, 7, and 8 are not located within the tunnel and can be considered to be in a good real-time dynamic signal state. Vehicle 2 is adjacent to vehicle 3, which belongs to the first category, and vehicle 7 is adjacent to vehicle 6, which also belongs to the first category. Vehicles 1 and 8 are not adjacent to any first category vehicles. Therefore, vehicles 2 and 7 can be determined to belong to the second category, while vehicles 1 and 8 belong to other categories other than the first and second categories. Based on this, the vehicles in the first and second categories are vehicles 2-7, and the relative positions between vehicles 2-7 include the relative positions between vehicles 2 and 3, the relative positions between vehicles 3 and 4, the relative positions between vehicles 4 and 5, the relative positions between vehicles 5 and 6, and the relative positions between vehicles 6 and 7.

[0039] It should be noted that the first category of vehicles and the second category of vehicles in the embodiment of the present application may include one vehicle or multiple vehicles. In addition, a vehicle may belong to the first category of vehicles in a period of time and may belong to the second category of vehicles in another period of time, for example, Figure 3 As shown, when the vehicle 6 is in the tunnel, the vehicle 6 belongs to the first category of vehicles, and after the vehicle 6 exits the tunnel, the vehicle 6 may belong to the second category of vehicles.

[0040] The vehicle can obtain its own position and real-time dynamic signals in real time through the inertial navigation module installed thereon, and can sense the relative position between the vehicle and its neighboring vehicles through the sensing device (such as a camera) thereon, and upload the obtained own position, real-time dynamic signal and relative position to the server. The server can receive its own position, real-time dynamic signal and relative position uploaded by multiple vehicles. Among them, the vehicle and the server can communicate based on the vehicle identification (Identity Document, referred to as ID). It should be noted that the inertial navigation modules and sensing devices deployed by different vehicles may be different, so that the methods adopted by the inertial navigation modules of different vehicles to obtain the vehicle's own position and real-time dynamic signals may be different, and the methods adopted by the sensing devices of different vehicles to obtain the relative position may be different. Therefore, the specific method for how the vehicle obtains its own position, real-time dynamic signal and relative position can be referred to the inertial navigation module and sensing device of the specific vehicle, which will not be described in detail here.

[0041] Based on real-time dynamic signals uploaded by multiple vehicles, the real-time dynamic signal status of each of the multiple vehicles can be determined. The real-time dynamic signal status includes the aforementioned weak real-time dynamic signal status and the aforementioned good real-time dynamic signal status. For example, a status bit of the real-time dynamic signal of each of the multiple vehicles can be obtained. If the status bit indicates that the solution state of the real-time dynamic signal is a fixed solution state, the real-time dynamic signal status of the vehicle corresponding to the status bit is determined to be a good real-time dynamic signal state. If the status bit indicates that the solution state of the real-time dynamic signal is not a fixed solution state, the real-time dynamic signal status of the vehicle corresponding to the status bit is determined to be a weak real-time dynamic signal state. After determining the real-time dynamic signal status of the vehicle, a vehicle with a weak real-time dynamic signal status can be determined as a first-category vehicle; a vehicle with a good real-time dynamic signal status adjacent to a first-category vehicle can be determined as a second-category vehicle. In other words, when receiving a real-time dynamic signal uploaded by any vehicle, the vehicle can be determined to be a first-category vehicle or a second-category vehicle based on the solution state of the status bit of the real-time dynamic signal.

[0042] Among them, the solution state of the real-time dynamic signal can characterize the solution accuracy of the real-time dynamic signal. In the order of decreasing solution accuracy of the real-time dynamic signal, the solution state of the real-time dynamic signal can include a fixed solution state, a floating-point solution state (also known as a differential solution state) and a single-point solution state. Among them, the solution accuracy of the real-time dynamic signal corresponding to the fixed solution state is at the centimeter level, usually less than or equal to 3 centimeters. It can also be understood that in the fixed solution state, the distance between the vehicle coordinates calculated based on the real-time dynamic signal and the actual vehicle coordinates is usually less than or equal to 3 centimeters. The solution accuracy of the real-time dynamic signal corresponding to the floating-point solution state is between the centimeter level and the meter level, usually greater than 3 centimeters and less than or equal to 3 meters. It can also be understood that in the floating-point solution state, the distance between the vehicle coordinates calculated based on the real-time dynamic signal and the actual vehicle coordinates is usually greater than 3 centimeters and less than or equal to 3 meters. The calculation accuracy of the real-time dynamic signal corresponding to the single-point solution state is at the meter level, usually greater than 3 meters and less than or equal to 10 meters. It can also be understood that in the single-point solution state, the distance between the vehicle coordinates calculated based on the real-time dynamic signal and the actual vehicle coordinates is usually greater than 3 meters and less than or equal to 10 meters.

[0043] In some embodiments, the implementation process of "determining a vehicle whose real-time dynamic signal status is a good real-time dynamic signal status and is adjacent to a first-category vehicle as a second-category vehicle" is as follows: taking the first-category vehicle as the central vehicle, the real-time dynamic signal status of the adjacent vehicles of the first-category vehicle is obtained. If the real-time dynamic signal status of the adjacent vehicle is a good real-time dynamic signal status (i.e., the real-time dynamic signal of the adjacent vehicle is good), the adjacent vehicle in the good real-time dynamic signal status is determined to be a second-category vehicle. If the real-time dynamic signal status of the adjacent vehicle is a weak real-time dynamic signal status (i.e., the real-time dynamic signal of the adjacent vehicle is relatively weak), the adjacent vehicle in the weak real-time dynamic signal status is taken as the new central vehicle, and the real-time dynamic signal status of the adjacent vehicles of the new central vehicle is obtained, until all the adjacent vehicles of the central vehicle are in a good real-time dynamic signal status.

[0044] As an example, see Figure 3 , assuming that vehicle 5 is the first to be detected in a weak real-time dynamic signal state, then with vehicle 5 as the center vehicle, the real-time dynamic signal states of vehicle 5's adjacent vehicles (i.e., vehicles 4 and 6) are acquired. If vehicles 4 and 6 are still in a weak real-time dynamic signal state, then with vehicles 4 and 6 as the new center vehicles, the real-time dynamic signal states of vehicles 4 and 6's adjacent vehicles (i.e., vehicles 3, 5, and 7) are acquired. Since the real-time dynamic signal state of vehicle 5 is already known, there is no need to re-acquire the real-time dynamic signal state of vehicle 5, nor is there any need to re-judge the real-time dynamic signal state of vehicle 5. Only the real-time dynamic signal states of vehicles 3 and 7 need to be determined. Figure 3 As shown, the real-time dynamic signal of vehicle 7 is good and is in a good real-time dynamic signal state. At this time, vehicle 7 can be determined to be a second-category vehicle, and the search for adjacent vehicles of vehicle 7 is no longer required. However, vehicle 3 is in a weak real-time dynamic signal state. At this time, it is still necessary to use vehicle 3 as the new center vehicle to obtain the real-time dynamic signal state of vehicle 3's adjacent vehicles (vehicles 2 and 4). Since the real-time dynamic signal state of vehicle 4 is known, there is no need to re-acquire the real-time dynamic signal state of vehicle 4, nor is there any need to re-judge the real-time dynamic signal state of vehicle 4. Only the real-time dynamic signal state of vehicle 2 needs to be judged. Figure 3 As shown, the real-time dynamic signal of vehicle 2 is good and is in a good real-time dynamic signal state. At this time, it can be determined that vehicle 2 is a second-category vehicle, and the adjacent vehicles of vehicle 2 are no longer searched.

[0045] In step S120, the self-posture of the first type of vehicle and the self-posture of the second type of vehicle are used as variables of the factor graph, and the relative posture uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signal of the first type of vehicle and the real-time dynamic signal of the second type of vehicle are used as factors of the factor graph to form a factor graph. The factor graph is an estimation model for solving the probability of the variable for a given factor.

[0046] Among them, the factor graph can refer to a bipartite graph composed of factors connecting variables. Variables represent unknown random variables in the estimation problem. Factors represent probabilistic constraints on variables, and the probabilistic constraints come from measurements or prior knowledge. In an embodiment of the present application, the self-position of the first type of vehicle and the self-position of the second type of vehicle are variables of the factor graph, represented as variable vertices. The relative position between the vehicles in the first type of vehicle and the second type of vehicle and the real-time dynamic signals of each vehicle in the first type of vehicle and the second type of vehicle are factors of the factor graph, which are probabilistic constraints on variables, wherein the relative position between the vehicles in the first type of vehicle and the second type of vehicle is represented as an edge connecting the variable vertices, and the real-time dynamic signals of each vehicle in the first type of vehicle and the second type of vehicle are represented as factor vertices of the factor graph, and the factor vertices are connected to the variable vertices.

[0047] As an example, a factor graph can be Figure 4 As shown, the blank circles represent the self-poses uploaded by the vehicles in the first and second categories, respectively. The circles filled with diagonal lines represent the real-time dynamic signals of the vehicles in the first and second categories. The dotted arrows between the blank circles represent the relative poses between the vehicles calculated by the cameras on the vehicles. Specifically, the dotted arrows represent the relative poses between the vehicles corresponding to the two connected blank circles. The relative pose corresponding to each dotted arrow can be the relative pose uploaded by the vehicle corresponding to the blank circle pointing to the dotted arrow. For example, the dotted arrow from blank circle 1 to blank circle 5 represents the relative pose between vehicle 1 and vehicle 5 uploaded by vehicle 1. The solid arrow between the blank circle and the diagonal filled circle indicates that the blank circle and the diagonal filled circle correspond to the same vehicle. That is, the self-poses of the vehicles represented by the blank circles connected by the same solid arrow and the real-time dynamic signals represented by the diagonal filled circle are uploaded by the same vehicle. For example, the self-pose of vehicle 5 corresponding to blank circle 5 and the real-time dynamic signal of vehicle 5 corresponding to diagonal filled circle 5 are uploaded by the same vehicle 5.

[0048] In some embodiments, the information carried by the real-time dynamic signal in the factor graph includes the vehicle posture obtained by positioning the vehicle according to the real-time dynamic signal (hereinafter referred to as the real-time dynamic posture of the vehicle) and the covariance matrix of the real-time dynamic signal, and the covariance matrix can be equivalent to the inverse matrix of the confidence of the real-time dynamic signal (confidence can be understood as weight). The factor graph can be optimized according to the real-time dynamic posture and the confidence of the real-time dynamic signal. Generally speaking, the higher the confidence of the real-time dynamic signal, the stronger the signal strength of the real-time dynamic signal, and the lower the confidence of the real-time dynamic signal, the lower the signal strength of the real-time dynamic signal.

[0049] It should be noted that after all first-category and second-category vehicles are determined, the first-category vehicle's own position, the second-category vehicle's own position, the relative position between vehicles in the first and second categories, and the real-time dynamic signal of each vehicle in the first and second categories can be added to the factor graph. Alternatively, when first-category vehicles are determined, that is, as long as a vehicle is in a weak real-time dynamic signal state, the vehicle's own position and real-time dynamic signal, the vehicle's neighboring vehicles' own position and real-time dynamic signal, and the relative position between the vehicle and its neighboring vehicles can be added to the factor graph. With the vehicle added to the factor graph as the central vehicle, the center vehicle's neighboring vehicles' own position and real-time dynamic signal, as well as the relative position between the center vehicle and its neighboring vehicles, can be added outwards until all of the center vehicle's neighbors are in a good real-time dynamic signal state, that is, when all of the center vehicle's neighbors have good real-time dynamic signals, no further data will be added to the factor graph. For a detailed description, please refer to the relevant section above and will not be repeated here.

[0050] Step S130: Optimize the factor graph to obtain an optimized posture of each vehicle in the first category and the second category.

[0051] Optimizing the factor graph may refer to adjusting the values ​​of the variables in the factor graph so as to maximize the continuous product of all factors. The posterior probability corresponding to the maximum continuous product of all factors is the maximum posterior probability, where continuous product means continuous product, and the symbol "∏" is usually used to represent continuous product. Wherein, the posterior probability refers to the possibility that something has happened and the reason for the occurrence of this thing is caused by a certain factor. The variables in the embodiment of the present application refer to the own posture of the first type of vehicle and the own posture of the second type of vehicle, and the factors of the variables refer to the relative posture between the first type of vehicle and the second type of vehicle uploaded by the first type of vehicle and the second type of vehicle, as well as the real-time dynamic signals of each vehicle in the first type of vehicle and the second type of vehicle. Wherein, the relative posture between the vehicles in the first type of vehicle and the second type of vehicle is a binary factor, because one relative posture is associated with two variables (i.e., the own posture of the two vehicles), and the real-time dynamic signals of the first type of vehicle and the second type of vehicle are unary factors, and one real-time dynamic signal is associated with one variable (i.e., the own posture of the vehicle). In factor graph optimization, when new variables and factors are added, it is first necessary to analyze the connection and influence relationship between the newly added variables and factors and the factor graph, and consider which previously stored information can continue to be used and which must be recalculated. Finally, only the variables associated with the newly added variables and factors are optimized.

[0052] In some embodiments, an optimization algorithm can be used to adjust the self-positions of the first and second class vehicles, and the posterior probability after each adjustment is calculated until a maximum posterior probability is obtained, wherein the posterior probability corresponding to the self-positions of the first and second class vehicles is equal to the product of various factors (the relative position uploaded by the first and second class vehicles, the real-time dynamic signal of the first class vehicle, and the real-time dynamic signal of the second class vehicle), and the position corresponding to the maximum posterior probability is the optimized position of each of the first and second class vehicles. The optimization algorithm may include, but is not limited to, a Gauss-Newton (GN) algorithm and a Levenberg-Marquardt (LM) algorithm.

[0053] For example, here we combine Figure 4 The factor graph shown illustrates how factor graph optimization is performed. Figure 4 The factor graph shown can be expressed as follows:

[0054] φ(X)=φ(x1)*φ(x2)*φ(x3)*φ(x4)*φ(x5)*φ(y 1→5 )*φ(y 5→1 )*φ(y 1→2 )*φ(y 2→1 )*φ(y 1→3 )*φ(y 3→1 )*φ(y 5→2 )*φ(y 2→5 )*φ(y 2→3 )*φ(y 3→2 )*φ(y 2→4 )*φ(y 4→2 )*φ(y 3→4 )*φ(y 4→3 )*φ(X1)

[0055] Among them, φ(X) represents the product of all factors, that is, the posterior probability corresponding to the self-position of the first type of vehicle and the second type of vehicle; φ(x1) represents the real-time dynamic signal of vehicle 1; φ(x2) represents the real-time dynamic signal of vehicle 2; φ(x3) represents the real-time dynamic signal of vehicle 3; φ(x4) represents the real-time dynamic signal of vehicle 4; φ(x5) represents the real-time dynamic signal of vehicle 5; φ(y 1→5 ) represents the relative position between vehicle 1 and vehicle 5 uploaded by vehicle 1; φ(y 5→1 ) represents the relative position between vehicle 1 and vehicle 5 uploaded by vehicle 5; φ(y 1→2 ) represents the relative position between vehicle 1 and vehicle 2 uploaded by vehicle 1; φ(y 2→1 ) represents the relative position between vehicle 1 and vehicle 2 uploaded by vehicle 2; φ(y1→3 ) represents the relative position between vehicle 1 and vehicle 3 uploaded by vehicle 1; φ(y 3→1 ) represents the relative position between vehicle 1 and vehicle 3 uploaded by vehicle 3; φ(y 2→5 ) represents the relative position between vehicle 2 and vehicle 5 uploaded by vehicle 2; φ(y 5→2 ) represents the relative position between vehicle 2 and vehicle 5 uploaded by vehicle 5; φ(y 2→3 ) represents the relative position between vehicle 2 and vehicle 3 uploaded by vehicle 2; φ(y 3→2 ) represents the relative position between vehicle 2 and vehicle 3 uploaded by vehicle 3; φ(y 2→4 ) represents the relative position between vehicle 2 and vehicle 4 uploaded by vehicle 2; φ(y 4→2 ) represents the relative position between vehicle 2 and vehicle 4 uploaded by vehicle 4; φ(y 3→4 ) represents the relative position between vehicle 3 and vehicle 4 uploaded by vehicle 3; φ(y 4→3 ) represents the relative position between vehicle 3 and vehicle 4 uploaded by vehicle 4; φ(X1) represents the prior factor, that is, the posterior probability calculated after the previous adjustment of the position of the first type of vehicle and the second type of vehicle. When the above expression is calculated for the first time, φ(X1) can be a preset value, so that the above expression can be solved.

[0056] Based on the above expression, optimizing the factor graph is to find the maximum value of the product of all factors of the factor graph, that is, the following expression can be used to express the optimization of the factor graph. Figure 4 The factor graph shown is optimized:

[0057]

[0058] Among them, X * The variable X corresponding to the maximum value of the product of all factors of the factor graph (i.e., the maximum posterior probability); the argmax(φ(X)) function is used to solve the variable corresponding to the maximum value of φ(X); arg max X Π i φ i (X i ) represents the variable corresponding to the maximum value (maximum posterior probability) of the product of all factors of the factor graph after the i-th adjustment.

[0059] As shown in the following expression, assuming all factors are negative exponential functions and taking the negative logarithm of the function, the negative exponential function maximization problem is actually transformed into a nonlinear least squares problem. Commonly used methods are iterative methods, such as the Gauss-Newton method.

[0060]

[0061] Taking the Gauss-Newton method as an example, as shown in the following expression, given an initial value, a set of possible system state variables (for example, the factors of the factor graph in the embodiment of the present application, that is, the relative posture uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signal of the first type of vehicle, and the real-time dynamic signal of the second type of vehicle) are used to solve a modification value to make the least squares value as small as possible, and then the value is added to the original value (that is, the initial value is modified), and then substituted back into the original function, and the initial value is iteratively adjusted. After reaching a certain stopping condition, the optimal modification value is obtained.

[0062]

[0063]

[0064] △X * =X0+△X

[0065] The intermediate incremental solution step is typically performed by linearizing the nonlinear function (performing a first-order Taylor expansion on the nonlinear function) to obtain a linear least squares problem. For example, the normal equation method and the orthogonal triangular decomposition method (also known as QR decomposition) can be used.

[0066] It should be noted that the vehicle posture optimized by the server includes the optimized vehicle posture of each vehicle in the first category of vehicles and the second category of vehicles. Data exchange can be achieved between the server and the vehicle based on the vehicle identification to transmit the vehicle posture optimized by the server. Usually, the first category of vehicles will use the vehicle posture optimized by the server for posture correction due to poor real-time dynamic signals, while the second category of vehicles can choose to use the vehicle posture optimized by the server for posture correction according to actual needs due to good real-time dynamic signals, or directly use the vehicle posture obtained by the global satellite positioning system of the vehicle for posture correction. This application does not impose specific restrictions on whether the second category of vehicles use the vehicle posture optimized by the server for posture correction.

[0067] The embodiments of the present application provide Figure 2The vehicle posture optimization method shown can use the self-posture of the first type of vehicle and the self-posture of the second type of vehicle as variables of the factor graph, use the relative posture uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signal of the first type of vehicle and the real-time dynamic signal of the second type of vehicle as factors of the factor graph, form a factor graph, and optimize the posture of each vehicle in the first type of vehicle and the second type of vehicle. This can provide an optimized posture for the first type of vehicle with weak real-time dynamic signal, so that the first type of vehicle can use the optimized self-posture to perform posture correction, thereby improving the positioning accuracy of the vehicle in the weak real-time dynamic signal scenario and improving the accuracy of vehicle navigation. In addition, the vehicle posture optimization method can be implemented through the equipment already installed on the vehicle, without the need to add expensive sensing equipment (for example, the above-mentioned multi-line laser radar). In other words, the vehicle posture optimization method can take into account both the positioning accuracy and cost of the vehicle in the weak real-time dynamic signal scenario.

[0068] Figure 5 1 is a flow chart of a vehicle posture correction method provided in one embodiment of the present application. The vehicle posture correction method can be applied to a vehicle or a vehicle posture correction device. The vehicle posture correction method can include the following steps S210 to S230.

[0069] In step S210, the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its adjacent vehicles are obtained and uploaded to the server, so that the server can optimize the vehicle posture according to the vehicle posture optimization method provided in the embodiment of the present application.

[0070] As mentioned above, the vehicle can obtain the vehicle's own posture and real-time dynamic signal in real time through the inertial navigation module installed thereon, and can perceive the relative posture between the vehicle and its adjacent vehicles through the sensing device (such as a camera) thereon. It should be noted that the data used to calculate the vehicle's own posture in different real-time dynamic signal states are different. For example, a vehicle in a weak real-time dynamic signal state uses inertial navigation data to calculate its own posture, while a vehicle in a good real-time dynamic signal state uses inertial navigation data and real-time dynamic data to calculate its own posture. Among them, inertial navigation data can refer to data obtained by the inertial measurement unit in the vehicle's inertial navigation module. Real-time dynamic data can refer to data obtained by parsing the real-time dynamic signal obtained by the global satellite navigation system in the vehicle's inertial navigation module.

[0071] In some embodiments, when the ego vehicle is in a weak real-time dynamic signal state, the ego vehicle's inertial navigation data can be fused through a State Error Kalman Filter (SEKF) to predict the vehicle's own posture, thereby obtaining the vehicle's own posture.

[0072] In some embodiments, when the vehicle is in a good state of real-time dynamic signal, the vehicle's inertial navigation data and real-time dynamic data can be fused through a state error Kalman filter to predict the vehicle's own posture, thereby obtaining the fused own posture.

[0073] It should be noted that the method by which a vehicle determines whether its real-time dynamic signal state is a weak real-time dynamic signal state or a good real-time dynamic signal state is the same as the method by which the server determines the real-time dynamic signal state of the vehicle. Both methods are based on the status bit of the vehicle's real-time dynamic signal. For the specific process, please refer to the part about determining the real-time dynamic signal state of the vehicle in step S110 above, which will not be repeated here. In some embodiments, after the vehicle determines its real-time dynamic signal state, it can directly upload its real-time dynamic signal state together with the real-time dynamic signal to the server. In this way, the server does not need to determine the real-time dynamic signal state of the vehicle, which can save vehicle posture optimization time.

[0074] The state error Kalman filter may include a vehicle posture prediction module and a posture correction module. The vehicle posture prediction module is used to predict the vehicle's own posture, and the posture correction module is used to correct the predicted own posture. The embodiment of the present application is set in a weak real-time dynamic signal scenario, and the optimized vehicle posture sent by the server is used to correct the predicted own posture. In a scenario with good real-time dynamic signals, the vehicle posture obtained by the global satellite navigation system is used to correct the predicted own posture.

[0075] The error state transfer process of the state error Kalman filter is shown in the following expression:

[0076] δp←δp+δvΔt

[0077] δv←δv+(-R[a m -a b ]×δθ-Rδa b +δg)△t+v i δθ←R T {(w m -w b )△t}δθ-δw b Δt+θ i

[0078] δa b ←δa b +a i

[0079] δw b ←w b +w i

[0080] Among them, δp represents the vehicle position error; δv represents the vehicle speed error; Δt represents the time error; R represents the rotation matrix; a m Represents the accelerometer measurement value; a b Indicates the accelerometer bias; δa b represents the error of accelerometer bias; δθ represents the vehicle heading angle error; δg represents the error of vehicle gravity acceleration; v i represents the vehicle speed modeled by a white Gaussian process; R T represents the transposed matrix of the rotation matrix R; w m Indicates the gyroscope measurement value; w b Indicates the gyroscope deviation; δw b Indicates the gyroscope deviation w b The error of θ i represents the vehicle heading angle modeled by the white Gaussian process; a i represents the bias estimate of the accelerometer modeled by a white Gaussian process; w i represents the bias estimate of the gyroscope modeled by a white Gaussian process.

[0081] In some embodiments, the vehicle's camera can sense adjacent vehicles to obtain the vehicle's neighboring vehicles. The relative pose between the vehicle and its neighboring vehicles can be determined based on the camera's intrinsic and extrinsic parameters. Camera intrinsic parameters refer to parameters related to the camera's own characteristics, such as the camera's focal length and pixel size. Camera extrinsic parameters refer to parameters of the camera in the world coordinate system, such as the camera's position and rotation. The rotation matrix and translation matrix together describe how to transform a point from the world coordinate system to the camera coordinate system. The rotation matrix describes the orientation of the coordinate axes of the world coordinate system relative to the camera coordinate axes. The translation matrix describes the position of the spatial origin in the camera coordinate system. The device's intrinsic and extrinsic parameters are typically calibrated and stored in the camera's flash memory at the time of shipment. Therefore, the camera's intrinsic and extrinsic parameters can be obtained from the camera's flash memory. Objects surrounding the vehicle (e.g., a vehicle) can be transformed to the camera coordinate system using the intrinsic parameters, and then to the vehicle coordinate system using the extrinsic parameters. This allows the relative pose of the vehicle's neighboring vehicles relative to the vehicle to be determined. In particular, if the camera's intrinsic and extrinsic parameters do not exist in the camera's flash memory, the Zhang Zhengyou calibration method can be used to calculate the camera's intrinsic and extrinsic parameters.

[0082] The vehicle can upload its own posture predicted by the state error Kalman filter and the relative posture between the vehicle and its adjacent vehicles perceived by the camera to the server, so that the server can optimize the vehicle posture using the vehicle posture optimization method provided in the embodiment of the present application.

[0083] Step S220: When the vehicle is in a weak real-time dynamic signal state, obtain the optimized vehicle posture sent by the server.

[0084] As previously mentioned, since the server-optimized vehicle pose includes the optimized pose for vehicles in weak real-time dynamic signal conditions, when the ego vehicle is in a weak real-time dynamic signal condition, it can request the optimized pose from the server based on its own vehicle ID. The server can then send the optimized pose to the ego vehicle based on its own vehicle ID. After obtaining the optimized pose, the ego vehicle can perform pose corrections based on the optimized pose, improving positioning accuracy in weak real-time dynamic signal conditions and, consequently, improving navigation accuracy.

[0085] Step S230 , correcting the vehicle's posture according to the optimized vehicle's posture.

[0086] The optimized ego-vehicle posture is used as the observation value of the state error Kalman filter, and the state error Kalman filter is used to correct the ego-vehicle posture. The specific method of using the state error Kalman filter to correct the vehicle posture can be found in related technologies and will not be described here to save space.

[0087] The embodiments of the present application provide Figure 5 The vehicle posture correction method shown is that when the vehicle is in a weak real-time dynamic signal scenario, the server can optimize the posture of each vehicle in the first and second categories based on the posture of the first type of vehicle in a weak real-time dynamic signal state, the posture of the second type of vehicle in a good real-time dynamic signal state and adjacent to the first type of vehicle, the relative posture between the vehicles in the first and second categories, and the real-time dynamic signal of each vehicle in the first and second categories. After that, the optimized posture of the vehicle is obtained from the server, and the posture of the vehicle is corrected according to the optimized posture of the vehicle, thereby improving the positioning accuracy of the vehicle in the weak real-time dynamic signal scenario and improving the accuracy of vehicle navigation. In addition, the method can be implemented by the equipment already installed on the vehicle without the need for adding expensive sensing equipment (for example, the above-mentioned multi-line laser radar). In other words, the vehicle posture optimization method can simultaneously take into account the positioning accuracy and cost of the vehicle in the weak real-time dynamic signal scenario.

[0088] Figure 6 1 is a flow chart of a vehicle posture correction method provided in one embodiment of the present application. The vehicle posture correction method can be applied to a vehicle or a vehicle posture correction device. The vehicle posture correction method can include the following steps S310 to S330.

[0089] In step S310, the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its adjacent vehicles are obtained and uploaded to the server, so that the server can optimize the vehicle posture using the vehicle posture optimization method provided in the embodiment of the present application.

[0090] Step S320: When the ego vehicle is in a weak real-time dynamic signal state, the ego vehicle's inertial navigation data is fused through a state error Kalman filter to obtain a fused ego vehicle posture; the optimized ego vehicle posture sent by the server is obtained; and the fused ego vehicle posture is corrected according to the optimized ego vehicle posture.

[0091] The detailed description of step S310 and step S320 can refer to the above-mentioned steps S210 to S230 and will not be repeated here.

[0092] Step S330: When the ego vehicle is in a good state of real-time dynamic signal, the ego vehicle's inertial navigation data and real-time dynamic data are fused through a state error Kalman filter to obtain a fused ego vehicle posture, and the fused ego vehicle posture is corrected according to the ego vehicle posture obtained by the global satellite navigation system.

[0093] When the ego vehicle is in a good state of real-time dynamic signal, the ego vehicle posture obtained by the global satellite navigation system can be used as the observation value of the state error Kalman filter to correct the ego vehicle posture.

[0094] Compared to Figure 5 The vehicle posture correction method shown in the embodiment of the present application provides Figure 6 The vehicle posture correction method shown also has the following technical effects: when the vehicle is in a scenario with weak real-time dynamic signal, the vehicle posture optimized by the server is used to correct the vehicle posture; when the vehicle is in a scenario with good real-time dynamic signal, the vehicle posture obtained by the global satellite navigation system is used to correct the vehicle posture. Different posture correction methods can be adopted for different application scenarios, so that the vehicle can be suitable for various scenarios, and the availability and positioning accuracy of the vehicle's positioning function are improved.

[0095] Figure 7 1 is a timing flow diagram of a vehicle posture correction method provided in one embodiment of the present application. The vehicle posture correction method can be applied to a vehicle posture correction system. The vehicle posture correction method can include the following steps 1 to 6.

[0096] Step 1: Multiple vehicles upload their own positions, real-time dynamic signals and relative positions to the server. Figure 7As shown, vehicle 1 can upload its own position, real-time dynamic signals, and the relative position between vehicle 1 and its adjacent vehicles to the server, and vehicle 2 can upload its own position, real-time dynamic signals, and the relative position between vehicle 2 and its adjacent vehicles to the server.

[0097] In step 2, the server obtains the status bit of the real-time dynamic signal of each of the multiple vehicles and determines whether the vehicle is a first-category vehicle or a second-category vehicle based on the status bit of the real-time dynamic signal of each vehicle. For a detailed description of step 2, please refer to the corresponding portion of step S110 and will not be repeated here.

[0098] In step three, the server uses the self-position of the first type of vehicle and the self-position of the second type of vehicle as variables of the factor graph, and uses the relative position uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signal of the first type of vehicle and the real-time dynamic signal of the second type of vehicle as factors of the factor graph to form a factor graph. The factor graph is an estimation model for the probability of solving the variables given the factors.

[0099] In step 4, the server optimizes the factor graph to obtain the optimized posture of each vehicle in the first category and the second category.

[0100] Step 5: any vehicle in a weak real-time dynamic signal state among the multiple vehicles obtains the optimized vehicle posture sent by the server, corrects the vehicle posture according to the optimized vehicle posture, and outputs the corrected vehicle posture. Figure 7 As shown, when the vehicle 1 is in a weak real-time dynamic signal state, it can obtain the optimized posture of the vehicle 1 sent by the server, correct the posture of the vehicle 1 according to the optimized posture of the vehicle 1, and output the corrected posture of the vehicle 1. Figure 7 As shown, when vehicle 2 is in a weak real-time dynamic signal state, it can obtain the optimized posture of vehicle 2 sent by the server, correct the posture of vehicle 2 according to the optimized posture of vehicle 2, and output the corrected posture of vehicle 2.

[0101] Step 6: any vehicle in a good real-time dynamic signal state among the multiple vehicles corrects the vehicle posture according to the vehicle posture obtained by the global satellite navigation system and outputs the corrected vehicle posture. Figure 7 As shown, when the vehicle 1 is in a good state of real-time dynamic signal, the vehicle posture can be corrected according to the vehicle posture obtained by the global satellite navigation system, and the corrected vehicle posture is output. Figure 7 As shown, when the vehicle 2 is in a good state of real-time dynamic signal, the vehicle posture can be corrected according to the vehicle posture obtained by the global satellite navigation system, and the corrected vehicle posture can be output.

[0102] The embodiments of the present application provide Figure 7The vehicle posture correction method shown can improve vehicle positioning accuracy in weak real-time dynamic signal scenarios, thereby improving vehicle navigation accuracy. Furthermore, this method does not require the addition of expensive sensing equipment. In other words, this method can improve vehicle positioning accuracy in weak real-time dynamic signal scenarios using the vehicle's existing sensor and computing resources, achieving a balance between cost and vehicle positioning accuracy in weak real-time dynamic signal scenarios.

[0103] Figure 8 FIG2 is a schematic diagram of the structure of a vehicle posture optimization device provided in one embodiment of the present application. The vehicle posture optimization device 200 can be applied to a server. The vehicle posture optimization device 200 can include a data acquisition module 210, a data processing module 220, and a posture optimization module 230.

[0104] The data acquisition module 210 is used to obtain the self-position, real-time dynamic signal and relative position uploaded by multiple vehicles, wherein the relative position is the relative position between the vehicle uploading the relative position and the adjacent vehicles of the vehicle. The multiple vehicles include first-class vehicles and second-class vehicles. The first-class vehicles are in a weak real-time dynamic signal state, and the second-class vehicles are in a good real-time dynamic signal state and are adjacent to the first-class vehicles.

[0105] The data processing module 220 is used to use the own posture of the first type of vehicle and the own posture of the second type of vehicle as variables of the factor graph, and use the relative posture uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signal of the first type of vehicle and the real-time dynamic signal of the second type of vehicle as factors of the factor graph to form a factor graph; the factor graph is an estimation model for solving the probability of the variables given the factors.

[0106] The posture optimization module 230 is used to optimize the factor graph to obtain an optimized posture of each vehicle in the first category and the second category.

[0107] In some embodiments, the data processing module 220 is also used to use the own posture of the first type of vehicle and the own posture of the second type of vehicle as variables of the factor graph, and use the relative posture between the first type of vehicle and the second type of vehicle uploaded by the first type of vehicle and the second type of vehicle and the real-time dynamic signals of each vehicle in the first type of vehicle and the second type of vehicle as factors of the factor graph and add them into the factor graph.

[0108] In some embodiments, the posture optimization module 230 is also used to use the relative posture between the first type of vehicle and the second type of vehicles and the real-time dynamic signals of each vehicle in the first type of vehicle and the second type of vehicles as probabilistic constraints on the self-posture of the first type of vehicle and the self-posture of the second type of vehicle, and use an optimization algorithm to adjust the self-posture of the first type of vehicle and the second type of vehicle, and calculate the posterior probability after each adjustment until the maximum posterior probability is obtained. The posture corresponding to the maximum posterior probability is the optimized posture of each vehicle in the first type of vehicle and the second type of vehicle.

[0109] In some embodiments, the data acquisition module 210 is also used to determine the real-time dynamic signal status of each of the multiple vehicles based on the real-time dynamic signals uploaded by the multiple vehicles, the real-time dynamic signal status including a weak real-time dynamic signal status and a good real-time dynamic signal status; a vehicle with a weak real-time dynamic signal status is determined as a first category vehicle; a vehicle with a real-time dynamic signal status is determined as a good real-time dynamic signal status and adjacent to the first category vehicle as a second category vehicle.

[0110] In some embodiments, the data acquisition module 210 is also used to obtain the status bit of the real-time dynamic signal of each of the multiple vehicles; if the status bit represents that the solution state of the real-time dynamic signal is a fixed solution state, the real-time dynamic signal state of the vehicle corresponding to the status bit is determined to be a good real-time dynamic signal state; if the status bit represents that the solution state of the real-time dynamic signal is not a fixed solution state, the real-time dynamic signal state of the vehicle corresponding to the status bit is determined to be a weak real-time dynamic signal state.

[0111] Figure 9 FIG2 is a schematic diagram of the structure of a vehicle posture correction device provided in one embodiment of the present application. The vehicle posture correction device 300 can be applied to a vehicle. The vehicle posture correction device 300 can include a data upload module 310, a data acquisition module 320, and a posture correction module 330.

[0112] The data upload module 310 is used to obtain and upload the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its adjacent vehicles to the server, so that the server can optimize the vehicle posture according to the vehicle-connected posture optimization method provided in the embodiment of the present application.

[0113] The data acquisition module 320 is used to obtain the optimized vehicle posture sent by the server when the vehicle is in a weak real-time dynamic signal state.

[0114] The posture correction module 330 is used to correct the posture of the ego vehicle according to the optimized posture of the ego vehicle.

[0115] In some embodiments, the posture correction module 330 is further configured to use the optimized ego-vehicle posture as an observation value of a state error Kalman filter, and perform ego-vehicle posture correction through the state error Kalman filter.

[0116] In some embodiments, the data upload module 310 is further configured to fuse the inertial navigation data of the ego vehicle through a state error Kalman filter to obtain the ego vehicle's own position and posture when the ego vehicle is in a weak real-time dynamic signal state.

[0117] In some embodiments, the data uploading module 310 is further used to sense vehicles adjacent to the vehicle through the vehicle's camera to obtain the vehicle's adjacent vehicles; and determine the relative position between the vehicle and the vehicle's adjacent vehicles based on the camera's internal and external parameters.

[0118] In some embodiments, the data upload module 310 is further configured to fuse the vehicle's inertial navigation data and real-time dynamic data using a state error Kalman filter to obtain a fused self-position when the vehicle is in a good real-time dynamic signal state. The position correction module 330 is further configured to correct the fused self-position based on the vehicle's position acquired by the global satellite navigation system when the vehicle is in a good real-time dynamic signal state.

[0119] Those skilled in the art will clearly understand that the vehicle posture optimization device 200 provided in the embodiment of the present application can implement the vehicle posture optimization method provided in the embodiment of the present application, and the vehicle posture correction device 300 provided in the embodiment of the present application can implement the vehicle posture correction method provided in the embodiment of the present application. The specific working processes of the vehicle posture optimization device 200 and its modules and the vehicle posture correction device 300 and its modules can be referred to the corresponding processes of the corresponding methods in the embodiment of the present application, and will not be repeated here.

[0120] In the embodiments provided in the present application, the coupling, direct coupling or communication connection between the modules in the vehicle posture optimization device 200 and the vehicle posture correction device 300 shown or discussed may be indirect coupling or communication coupling through some interfaces, devices or modules, and may be electrical, mechanical or other forms, and the embodiments of the present application do not impose specific restrictions here.

[0121] In addition, the functional modules in the vehicle posture optimization device 200 and the vehicle posture correction device 300 can be integrated into a single processing module, or each module can exist physically separately, or two or more modules can be integrated into a single module. The above-mentioned integrated modules can be implemented in the form of hardware or software functional modules, and the embodiments of the present application do not impose specific limitations on this.

[0122] Figure 10This is a schematic diagram of the structure of the server provided in one embodiment of the present application. The server 400 can be connected to Figure 1 The server 400 may include one or more of the following components: a memory 410, one or more processors 420, and one or more applications, wherein the one or more applications may be stored in the memory 410 and used to, when called by the one or more processors 420, enable the server to execute the vehicle posture optimization method provided in the embodiments of the present application.

[0123] The processor 420 may include one or more processing cores. The processor 420 uses various interfaces and lines to connect various components within the entire server 400 and is used to run or execute instructions, programs, code sets, or instruction sets stored in the memory 410, as well as call and execute data stored in the memory 410, perform various functions of the server 400, and process data.

[0124] The processor 420 may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA).

[0125] Processor 420 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem handles wireless communications. It is understood that the modem may not be integrated into processor 420 and may be implemented separately via a communications chip.

[0126] The memory 410 may include a random access memory (RAM) or a read-only memory (ROM). The memory 410 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 410 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for implementing at least one function, instructions for implementing the embodiment of the above-mentioned vehicle posture optimization method, etc. The data storage area may store data created by the server 400 during use, etc.

[0127] Figure 11 FIG. 5 is a schematic diagram of the structure of a vehicle provided in an embodiment of the present application. Figure 1 The illustrated vehicles 110 may be identical. Vehicle 500 may include one or more of the following components: a memory 510, one or more processors 520, and one or more applications, wherein the one or more applications may be stored in the memory 510 and used to, when called by the one or more processors 520, cause a server to execute the vehicle posture correction method provided in an embodiment of the present application.

[0128] The processor 520 may include one or more processing cores. The processor 520 utilizes various interfaces and lines to connect to various components within the vehicle 500 and is configured to run or execute instructions, programs, code sets, or instruction sets stored in the memory 510, as well as call and execute data stored in the memory 510, perform various functions of the vehicle 500, and process data.

[0129] The processor 520 may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA).

[0130] The processor 520 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem handles wireless communications. It is understood that the modem may not be integrated into the processor 520 but may be implemented separately via a communications chip.

[0131] The memory 510 may include a random access memory (RAM) or a read-only memory (ROM). The memory 510 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 510 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for implementing at least one function, instructions for implementing the above-mentioned embodiment of the vehicle posture correction method, etc. The data storage area may store data created by the vehicle 500 during use, etc.

[0132] Figure 12 6 is a schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of the present application. The computer-readable storage medium 600 stores program code 610, which, when called by a processor, causes a server to execute the vehicle posture optimization method provided in an embodiment of the present application, or causes a vehicle to execute the vehicle posture correction method provided in an embodiment of the present application.

[0133] The computer-readable storage medium 600 may be, but is not limited to, an electronic memory such as a flash memory, an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a hard disk, or a read-only memory (ROM).

[0134] In some embodiments, the computer-readable storage medium 600 may include a non-transitory computer-readable storage medium (Non-TCRSM). The computer-readable storage medium 600 has storage space for program code 610 that executes any method step in the above method. These program codes 610 can be read from or written into one or more computer program products. The program code 610 can be compressed in an appropriate form.

[0135] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of this application and are not intended to limit them. Although this application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they may modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein with equivalents; and such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the various embodiments of this application.

Claims

1. A vehicle posture optimization method, characterized in that: include: Obtaining the self-position, real-time dynamic signal, and relative position uploaded by multiple vehicles, where the relative position is the relative position between the vehicle uploading the relative position and its adjacent vehicles, the multiple vehicles including first-category vehicles and second-category vehicles, the first-category vehicles having weak real-time dynamic signal conditions, and the second-category vehicles having good real-time dynamic signal conditions and being adjacent to the first-category vehicles; The first type of vehicle's own posture and the second type of vehicle's own posture are used as variables of a factor graph, and the relative postures uploaded by the first type of vehicle and the second type of vehicle, the real-time dynamic signals of the first type of vehicle, and the real-time dynamic signals of the second type of vehicle are used as factors of the factor graph to form a factor graph; the factor graph is an estimation model for solving the probability of the variables given the factors; The factor graph is optimized to obtain an optimized posture of each vehicle in the first category of vehicles and the second category of vehicles.

2. The method according to claim 1, characterized in that The posterior probability corresponding to the self-positions of the first type of vehicle and the second type of vehicle is equal to the product of each of the factors; The optimizing the factor graph to obtain the optimized posture of each vehicle in the first category and the second category includes: adjusting the posture of the first category and the second category using an optimization algorithm, and calculating the posterior probability after each adjustment until a maximum posterior probability is obtained, wherein the posture corresponding to the maximum posterior probability is the optimized posture of each vehicle in the first category and the second category.

3. The method according to claim 1 or 2, characterized in that After obtaining the self-positions, real-time dynamic signals, and relative positions uploaded by multiple vehicles, the method further includes: Determining a real-time dynamic signal state of each of the multiple vehicles based on the real-time dynamic signals uploaded by the multiple vehicles, the real-time dynamic signal state including a weak real-time dynamic signal state and a good real-time dynamic signal state; Determine a vehicle with a weak real-time dynamic signal state as a first-category vehicle; A vehicle whose real-time dynamic signal status is a good real-time dynamic signal status and is adjacent to the first-category vehicle is determined as a second-category vehicle.

4. The method according to claim 3, characterized in that The determining, based on the real-time dynamic signals uploaded by the multiple vehicles, the real-time dynamic signal status of each of the multiple vehicles includes: Obtaining a status bit of a real-time dynamic signal of each of the plurality of vehicles; If the state bit represents that the solution state of the real-time dynamic signal is a fixed solution state, determining that the real-time dynamic signal state of the vehicle corresponding to the state bit is a good state of the real-time dynamic signal; If the state bit represents that the solution state of the real-time dynamic signal is not a fixed solution state, it is determined that the real-time dynamic signal state of the vehicle corresponding to the state bit is a weak real-time dynamic signal state.

5. A vehicle posture correction method, characterized in that: include: Obtaining and uploading the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its neighboring vehicles to a server, so that the server optimizes the vehicle posture according to the method according to any one of claims 1 to 4; When the ego vehicle is in a weak real-time dynamic signal state, obtain the optimized ego vehicle posture sent by the server; The ego-vehicle posture is corrected according to the optimized ego-vehicle posture.

6. The method according to claim 5, characterized in that The obtaining of the vehicle's own posture includes: When the ego vehicle is in a weak real-time dynamic signal state, the ego vehicle's inertial navigation data is fused through the state error Kalman filter to obtain the ego vehicle's own position and posture.

7. A vehicle posture optimization device, characterized in that: include: a data acquisition module configured to acquire the self-position, real-time dynamic signal, and relative position uploaded by multiple vehicles, wherein the relative position is the relative position between the vehicle uploading the relative position and its adjacent vehicles, wherein the multiple vehicles include first-category vehicles and second-category vehicles, wherein the first-category vehicles are in a weak real-time dynamic signal state, and the second-category vehicles are in a good real-time dynamic signal state and are adjacent to the first-category vehicles; a data processing module, configured to use the self-positions of the first-category vehicle and the self-positions of the second-category vehicle as variables of a factor graph, and use the relative position uploaded by the first-category vehicle and the second-category vehicle, the real-time dynamic signals of the first-category vehicle, and the real-time dynamic signals of the second-category vehicle as factors of the factor graph, thereby forming a factor graph; the factor graph is an estimation model for solving the probability of the variables given the factors; A posture optimization module is used to optimize the factor graph to obtain an optimized posture of each vehicle in the first category and the second category.

8. A vehicle posture correction device, characterized in that: include: a data uploading module for acquiring and uploading the vehicle's own posture, real-time dynamic signals, and the relative posture between the vehicle and its neighboring vehicles to a server, so that the server can optimize the vehicle posture according to the method described in any one of claims 1 to 5; The data acquisition module is used to obtain the optimized vehicle posture sent by the server when the vehicle is in a weak real-time dynamic signal state; The posture correction module is used to correct the posture of the ego vehicle according to the optimized posture of the ego vehicle.

9. A server, characterized in that: include: Memory; one or more processors; One or more application programs, the one or more application programs being stored in the memory and configured to cause the server to execute the method according to any one of claims 1 to 4 when called by the one or more processors.

10. A vehicle, characterized in that: include: Memory; one or more processors; One or more applications, the one or more applications being stored in the memory and configured to cause the vehicle to execute the method according to any one of claims 5 to 6 when called by the one or more processors.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which is used to cause the server to execute the method according to any one of claims 1 to 4, or cause the vehicle to execute the method according to any one of claims 5 to 6 when called by one or more processors.

Images