Positioning method and device for autonomous driving of vehicles, vehicle, and storage medium

Abstract

This application relates to the technology of autonomous driving, and more particularly to a positioning method and apparatus, vehicle, and storage medium for autonomous driving of a vehicle. The method includes the steps of: obtaining a GDPR value for the vehicle's current positioning policy; calculating the current confidence level of the current positioning policy based on the GDPR value; determining that the current positioning policy is invalid if the current confidence level is lower than a reliable confidence level; replacing the current positioning policy with a candidate positioning policy; and obtaining the vehicle's current position based on the candidate positioning policy, based on the fusion of the vehicle's Global Navigation Satellite System (GNSS) data, LiDAR data, and inertial measurement data. This method effectively improves the accuracy of vehicle positioning when GNSS signals are relatively weak, thereby improving vehicle safety while also enhancing the user's driving experience.

Description

Technical Field

[0001] This application relates to the technical field of autonomous driving, and particularly relates to a positioning method, device, vehicle, and storage medium in the autonomous driving of vehicles.

Background Art

[0002] car Both can adopt a multi-sensor information fusion positioning technology that combines GNSS (Global Navigation Satellite System) and IMU (Inertial Measurement Unit), and by perceiving the surrounding environment, realize real-time positioning in a complex environment.

Summary of the Invention

[0003] Embodiments of this application ,car provide a positioning method and device, vehicle, and storage medium in the autonomous driving of both.

[0004] Embodiments of the first aspect of this application provide a positioning method in the autonomous driving of a vehicle. The method includes: obtaining a geometric dilution of precision GDOP value of the current positioning policy of the vehicle, where the GDOP value is used to indicate the positioning accuracy of the Global Navigation Satellite System GNSS data of the vehicle, and the current positioning policy is a GNSS positioning policy; calculating the current reliability of the current positioning policy based on the geometric dilution of precision GDOP value; when the current reliability is lower than a reliable reliability, determining that the current positioning policy is invalid, replacing the current positioning policy with a candidate positioning policy, and determining the current positioning of the vehicle based on the candidate positioning policy, where the candidate positioning policy is a positioning policy for obtaining the current positioning of the vehicle based on the fusion of the Global Navigation Satellite System GNSS data, lidar data, and inertial measurement data of the vehicle.

[0005] Selectively, in one embodiment of the present invention, obtaining the current position of the vehicle based on the fusion of the aforementioned GNSS data, LiDAR data, and inertial measurement data of the vehicle means determining the initial position and attitude information of the vehicle in the map coordinate system based on the GNSS data and the inertial measurement data, and determining the point cloud scene based on the initial position and attitude information and the LiDAR data. earth This includes generating a diagram and determining the current positioning of the vehicle based on the matching result between the point cloud data in the LiDAR data and the point cloud scene map.

[0006] Selectively, in one embodiment of the present invention, determining the initial position and attitude information of the vehicle in the map coordinate system based on the aforementioned GNSS data and inertial measurement data includes determining global positioning information in a reference coordinate system based on the GNSS data and inertial measurement data using an extended Kalman filtering method, and determining the initial position and attitude information of the vehicle in the map coordinate system based on the transformation relationship between the reference coordinate system and the map coordinate system, and the global positioning information.

[0007] Selectively, in one embodiment of the present invention, based on the above initial position and attitude information and the above lidar data, a point cloud scene earth Generating a figure includes constructing a point cloud map based on at least one keyframe from the LiDAR data corresponding to the initial position and orientation information; extracting keyframe data from the LiDAR data; extracting at least one point surface feature from the keyframe data, performing position and orientation matching on each point surface feature to obtain the position and orientation information of the keyframe; and inserting point cloud data into the point cloud map based on the position and orientation information of the keyframe to generate a point cloud scene map.

[0008] Selectively, in one embodiment of the present invention, before replacing the current positioning policy with the candidate positioning policy, the method further includes the steps of: obtaining the residual of the Euclidean distance between the point cloud data and the corresponding point in the point cloud scene map; calculating the confidence level of the candidate positioning policy based on the residual of the Euclidean distance; and, if the confidence level of the candidate positioning policy is equal to or greater than the current confidence level, performing an operation to replace the current positioning policy with the candidate positioning policy.

[0009] Selectively, in one embodiment of the present invention, after calculating the confidence level of the candidate positioning policy based on the residual of the Euclidean distance described above, the method further includes the step of indicating that the confidence level is too low if the confidence level of the candidate positioning policy is lower than the current confidence level.

[0010] Selectively, in one embodiment of the present invention, after calculating the confidence level of the candidate positioning policy based on the residual of the Euclidean distance described above, the method further includes the steps of determining whether the current confidence level is lower than the unreliable confidence level if the confidence level of the candidate positioning policy is lower than the current confidence level, and indicating that positioning is not possible if the current confidence level is lower than the unreliable confidence level.

[0011] Embodiments of a second aspect of the present invention provide a positioning device for autonomous driving of a vehicle, the device comprising: an acquisition module for acquiring a geometric precision degradation rate (GDOP) value of the vehicle's current positioning policy, the GDOP value being used to indicate the positioning precision of the vehicle's Global Navigation System (GNSS) data, and the current positioning policy being a GNSS positioning policy; a calculation module for calculating the current confidence level of the current positioning policy based on the geometric precision degradation rate (GDOP) value; and a processing module for determining that the current positioning policy is invalid if the current confidence level is lower than a reliable confidence level, replacing the current positioning policy with a candidate positioning policy, and determining the current positioning of the vehicle based on the candidate positioning policy, the candidate positioning policy being a positioning policy that obtains the current positioning of the vehicle based on the fusion of the vehicle's Global Navigation System (GNSS) data, lidar data, and inertial measurement data.

[0012] An embodiment of a third aspect of the present invention provides a vehicle including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor executes the program to realize a positioning method for the autonomous driving of the vehicle described in the embodiment.

[0013] A fourth embodiment of the present application provides a computer-readable storage medium that stores a computer program that, when executed by a processor, implements the positioning method for the autonomous driving of the vehicle described above. [Brief explanation of the drawing]

[0014] The above and / or additional aspects and advantages of the present application will be evident and easier to understand from the description of the embodiments based on the following drawings. [Figure 1] This is a flowchart of the positioning method for autonomous driving of a vehicle provided by the embodiment of the present invention. [Figure 2]This is a schematic diagram of edge point feature matching provided by the embodiment of the present application. [Figure 3] This is a schematic diagram of the planar point feature matching provided by the embodiment of the present application. [Figure 4] This is a schematic diagram illustrating the principle of switching between two positioning methods based on the calculated confidence provided by the embodiment of the present invention. [Figure 5] This is a schematic diagram of the structure of a positioning device for autonomous driving of a vehicle provided by an embodiment of the present invention. [Figure 6] This is a schematic diagram of the structure of a vehicle provided by an embodiment of the present invention. [Modes for carrying out the invention]

[0015] The embodiments of the present application will be described in detail below, and examples of the above embodiments are shown in the drawings, where the same or similar reference numerals throughout the text represent the same or similar elements, or elements having the same or similar functions. The embodiments described below with reference to the drawings are illustrative and are intended for use in interpreting the present application, and should not be understood as limiting the present application.

[0016] Before interpreting and explaining the embodiments of this application, we will first introduce the application scenarios of these embodiments.

[0017] One of the essential requirements for autonomous vehicles to travel on roads is to know their current position and orientation, as well as the target position and orientation. Therefore, sensors must be used to perceive the surrounding environment and enable real-time positioning of the vehicle's current position and orientation within that environment. Sensors commonly used for positioning include GNSS, LiDAR, and IMU. However, single sensors all have limitations in determining the vehicle's current position. Multi-sensor information fusion positioning technology overcomes the problems of single-sensor positioning technology, such as the limited amount of information available and its low timeliness and robustness. This effectively improves the accuracy and feasibility of positioning in complex environments. Consequently, multi-sensor information fusion positioning technology is a promising field of research in the current field of autonomous driving.

[0018] At present, the multi-sensor information fusion positioning technology generally includes a positioning technology that combines GNSS and IMU. The IMU can compensate for the defect that the update frequency of GNSS is low, and GNSS can correct the motion error of the IMU. However, in an underground tunnel or other places where the GNSS signal is low, GNSS may not be able to timely correct the motion error of the IMU. At this time, deviations or jumps may occur in the current positioning determined, and if this situation cannot be detected in a timely manner, it is easy to cause vehicle accidents.

[0019] Based on this, the present application provides a positioning method in the automatic driving of a vehicle. In this method, by obtaining the GDOP value of the current positioning policy of the vehicle, the current reliability of the current positioning policy is calculated. When the current reliability is lower than the reliable reliability, the current positioning policy is replaced with a candidate positioning policy, and the current positioning of the vehicle can be obtained based on the fusion of the GNSS data, lidar data and inertial measurement data of the vehicle, effectively improving the accuracy of vehicle positioning when the GNSS signal is relatively weak, thereby improving the safety of the vehicle and at the same time further satisfying the driving experience of the user.

[0020] Hereinafter, a positioning method and device in the automatic driving of a vehicle provided by an embodiment of the present application will be described with reference to the drawings.

[0021] FIG. 1 is a flowchart of a positioning method in the automatic driving of a vehicle provided by an embodiment of the present application. As shown in FIG. 1, the positioning method in the automatic driving of the vehicle includes the following steps.

[0022] In step S101, the GDOP value of the current positioning policy of the vehicle is obtained. The GDOP value is used to indicate the positioning accuracy of the GNSS data of the vehicle, and the current positioning policy is the GNSS positioning policy.

[0023] As can be understood, the embodiments of the present application can obtain the GDOP (geometric dilution precision) value of the current positioning policy of the vehicle. For example, based on PDOP (position dilution of precision) and TDOP (time dilution of precision) data, the GDOP value corresponding to the GNSS data can be calculated, and the calculated value can be used as the GDOP value of the current positioning policy, thereby effectively improving the feasibility of positioning in the automatic driving provided by the embodiments of the present application.

[0024] Among them, since PDOP is not only related to the quality of the observation results but also related to the geometric shape between the measured satellite and the receiver, PDOP may also be called the strength of accuracy. Exemplarily, PDOP can be calculated from HDOP (Horizontal Dilution of Precision) and VDOP (vertical dilution of precision).

[0025] In some embodiments, the GDOP calculation method can be represented by the following formula:

[0026]

Number

[0027] It should be noted that GDOP can represent the distance vector amplification factor between the receiver and the satellite due to the GNSS ranging error, and the volume of the shape depicted by the unit vector from the receiver to the satellite involved in the positioning solution is inversely proportional to GDOP. Therefore, GDOP can be called the geometric dilution of precision.

[0028] In step S102, the current confidence level of the current positioning policy is calculated based on the GDPR value.

[0029] To make it clear, an embodiment of the present invention can calculate the current confidence level of the current positioning policy based on GDP values. For example, the current confidence level can be calculated based on GDP values ​​to evaluate the accuracy of the position and orientation in the ENU (East-North-Up) coordinate system output by the current positioning policy.

[0030] This indicates that a larger GDOP value corresponds to lower GNSS data positioning accuracy, while a smaller GDOP value corresponds to higher GNSS data positioning accuracy. Therefore, the GDOP value determined in S101 shows a negative correlation with the current confidence level of the current positioning policy. In other words, a larger GDOP value corresponds to lower current confidence level of the current positioning policy, and a smaller GDOP value corresponds to higher current confidence level of the current positioning policy.

[0031] For example, the obtained GDPR values ​​can be normalized and then combined with a confidence measurement method. For instance, if the GDPR value is 0, the current confidence level is 1. By analogy, the current confidence level of the currently adopted GNSS positioning policy can be obtained as a criterion for evaluating the currently used GNSS positioning policy, thereby effectively improving the accuracy of positioning in autonomous driving.

[0032] In step S103, if the current confidence level is lower than the confidence level, it is determined that the current positioning policy is invalid, and the current positioning policy is replaced with a candidate positioning policy. Based on the candidate positioning policy, the vehicle's current positioning is determined, and the candidate positioning policy is a positioning policy that obtains the vehicle's current positioning based on the fusion of the vehicle's GNSS data, LiDAR data, and inertial measurement data.

[0033] Of these, the trustworthy level is the pre-set level of trust.

[0034] To make it clear, the embodiment of the present invention determines that the current positioning policy is invalid if the current confidence level is lower than the confidence level, and further replaces the current positioning policy with a candidate positioning policy, and determines the vehicle's current position based on the candidate positioning policy. After adopting the candidate positioning policy, the vehicle obtains its current position based on the fusion of the vehicle's GNSS data, LiDAR data, and inertial measurement data in the following steps, thereby effectively improving the accuracy of positioning in autonomous driving when the GNSS signal is relatively weak, and improving the quality of the map constructed in the autonomous driving scene.

[0035] In some embodiments, selectively, the implementation method for obtaining the current position of a vehicle based on the fusion of the vehicle's GNSS data, LiDAR data, and inertial measurement data may be as follows: Based on the GNSS data and inertial measurement data, the initial position and attitude information of the vehicle in the map coordinate system is determined, and based on the initial position and attitude information and LiDAR data, the point cloud scene earth A diagram is generated, and the vehicle's current positioning is determined based on the matching results between the point cloud data and the point cloud scene map in the LiDAR data.

[0036] In the embodiment of the present invention, first, the initial position and attitude information of a vehicle is determined based on GNSS data and inertial measurement data, and then the initial position and attitude information is corrected based on LiDAR data collected by the vehicle to obtain the vehicle's current position, thereby improving the accuracy of vehicle positioning when the GNSS signal is relatively weak.

[0037] Among these, the implementation method for determining the initial position and attitude information of a vehicle in a map coordinate system based on GNSS data and inertial measurement data is exemplified as follows: Using an extended Kalman filtering method, global positioning information in a reference coordinate system is determined based on GNSS data and inertial measurement data, and the initial position and attitude information of the vehicle in a map coordinate system is determined based on the transformation relationship between the reference coordinate system and the map coordinate system, and the global positioning information. That is, the global positioning information is transformed into a map coordinate system using coordinates to obtain the initial position and attitude information of the vehicle.

[0038] One feasible approach is that the GNSS and IMU in the embodiment of this application can be merged by an extended Kalman filtering method to output global positioning information in the ENU coordinate system. That is, the reference coordinate system is the ENU coordinate system, and the global positioning information is converted to a map coordinate system using coordinates, which is then used as initial position and attitude information for positioning using LiDAR data, thereby effectively improving the robustness of vehicle positioning.

[0039] For example, the embodiment of this application can complete the initialization of the positioning of an autonomous vehicle in a map coordinate system based on latitude and longitude information measured by GNSS in the ENU coordinate system and attitude angle information measured by IMU. The conversion relationship between the ENU coordinate system and the map coordinate system can be obtained using open-source tools, and their explanation is omitted here.

[0040] The selective determination of the vehicle's initial position and attitude information in the map coordinate system based on GNSS data and inertial measurement data may also be achieved by other methods, such as a strong tracking filtering algorithm, which will not be explained in detail here.

[0041] Furthermore, based on the initial position and orientation information and lidar data, the point cloud scene earth The implementation method for generating the map is as follows: A point cloud map is constructed based on at least one keyframe corresponding to the initial position and orientation information in the LiDAR data; keyframe data is extracted from the LiDAR data; at least one point surface feature is extracted from the keyframe data; position and orientation matching is performed on each point surface feature to obtain the position and orientation information of the keyframes; point cloud data is inserted into the map based on the position and orientation information of the keyframes to generate a point cloud scene map.

[0042] In one embodiment of the present invention, point cloud data scanned by a LiDAR on a vehicle while the vehicle is in the state of initial position and attitude information is used as at least one keyframe in the LiDAR data corresponding to the initial position and attitude information. In other words, a point cloud map is constructed based on the keyframe obtained from the initial position and attitude information.

[0043] Next, keyframe data is extracted from the lidar data. For example, keyframe data in the lidar data is extracted by setting translation thresholds and rotation thresholds for the lidar scan data. For instance, the translation threshold is set to a translation amount of 1m, and the rotation threshold is set to a position and attitude angle of 10 degrees.

[0044] After obtaining keyframe data, point and surface features are extracted based on the keyframe data, and positional and orientation matching is performed on each point and surface feature. Based on the obtained positional and orientation data, the point cloud data of the keyframes is inserted into a point cloud map, generating a point cloud scene map, thereby effectively reducing the amount of computation and lowering the computational power.

[0045] The embodiment of this application can calculate the positional transformation relationship by extracting features from point cloud data and performing feature matching. The extracted features include point features and planar features, and are abbreviated as point-plane features.

[0046] For example, when a new lidar scan data is received, edge points and plane points are calculated based on the average distance of 10 points before and after a single point on each scan line of the lidar scan data, and point features and plane features are further obtained. Of these, edge points are points on sharp edges in 3D space, with a relatively large difference in size between the edge point and surrounding points, relatively high curvature, and relatively high smoothness. Plane points are points on a smooth plane in 3D space, with a small difference in size between the plane point and surrounding points, relatively low curvature, and relatively low smoothness.

[0047] Illustratively, point feature matching is solved by minimizing the distance from an edge point to an edge line, and planar feature matching is solved by optimizing the distance from a planar point to a plane formed by adjacent planar points. For example, as shown in Figure 2, Figure 2 represents the point feature matching process corresponding to an edge point, where lines (1), (2), and (3) represent the scan lines of a multiline lidar. In the (k+1)th frame, edge point i lies on the scan line of line (2), and in the kth frame, edge points j and i on line (2) are closer together. When another closest edge point l is found on the adjacent line (1), l and j form an edge line. Therefore, for an edge feature point, the optimization goal is to minimize the distance from i to the line lj, and the feature of edge point i after the final optimization is taken as the extracted point feature.

[0048] Of these, the distance from edge point i to the edge line formed by edge points l and j is calculated as follows:

[0049]

number

[0050] For example, as shown in Figure 3, Figure 3 represents the matching process of planar features corresponding to planar points, where lines (1), (2), and (3) represent the scan lines of a multiline lidar. In the (k+1)th frame, planar point i lies on scan line (2). In the kth frame, planar points j and i on (2) are closer together. We find one more neighboring planar point l on the same line, and one more closest planar point m on the adjacent line (1). Points l, m, and j then form a plane. Therefore, for planar feature points, the optimization goal is to minimize the distance from planar point i to plane mlj. Finally, the features of planar point i after optimization are extracted as planar features.

[0051] Of these, the distance from plane point i to the plane formed by plane points m, l, and j is calculated as follows:

[0052]

number

[0053] In some embodiments, cumulative errors exist in the front-end laser odometer, so when an autonomous vehicle repeatedly travels the same area, a discrepancy occurs between the newly created map and the original map, causing a ghosting problem, which enables loopback detection. In the embodiment of the present invention, positioning data obtained by fusing GNSS and IMU, i.e., initial position and attitude information, is introduced as the initial value for loopback detection, the search range for loopback detection is set to within 10m of the initial value, the historical position closest to the current position is searched within the initial value range, and the current frame is matched using 10 keyframes before and after the historical position, and the keyframe that is closest in distance to the position and attitude corresponding to the current frame and has a longer interval time is found, thereby effectively accelerating the loopback detection speed and improving the accuracy of loopback detection.

[0054] It should be noted that the above embodiment exemplifies an implementation method for obtaining the current position of a vehicle based on the fusion of the vehicle's GNSS data, LiDAR data, and inertial measurement data. The embodiment of this application does not limit the implementation methods of the candidate positioning policy, and therefore, examples will not be given one by one here.

[0055] Selectively, in embodiments of the present invention, before replacing the current positioning policy with a candidate positioning policy, the residual Euclidean distance between the point cloud data and the corresponding point in the point cloud scene map may be obtained, the confidence level of the candidate positioning policy may be calculated based on the residual Euclidean distance, and if the confidence level of the candidate positioning policy is equal to or greater than the current confidence level, the operation to replace the current positioning policy with the candidate positioning policy may be performed, i.e., the current positioning policy may be allowed to be replaced with the candidate positioning policy.

[0056] In the embodiment of this invention, in order to establish a point cloud map, first the initialization of the vehicle's global position and orientation is completed based on GNSS and IMU, and then the point cloud data of the current frame and the point cloud scene are obtained. earth By matching with the figure, a position and orientation transformation matrix in the map coordinate system can be calculated, and the current position and orientation information of the autonomous vehicle can be obtained. Simultaneously, based on the calculated position and orientation transformation matrix, i.e., the rotation-translation matrix, the sum of residual Euclidean distances between the current LiDAR scan data, i.e., point cloud data, and corresponding points on the point cloud map is calculated, and the confidence level of the candidate positioning policy is characterized by this sum of residual Euclidean distances. The calculation method is as follows:

[0057]

number

[0058] For example, as shown in Figure 4, the confidence level of the current positioning policy, i.e., the GNSS positioning policy, is denoted as C2, and the confidence level of the candidate positioning policy, i.e., the Lidar+IMU+GNSS fusion positioning policy, is denoted as C1. If the calculated confidence level C1 is greater than or equal to the current confidence level C2, the system is permitted to switch to the candidate positioning policy, thereby effectively improving positioning accuracy and enhancing vehicle safety and reliability.

[0059] In some other embodiments, after calculating the confidence level of a candidate positioning policy, if the confidence level of the candidate positioning policy is lower than the current confidence level, positioning is performed using the current positioning policy as is, while simultaneously indicating that the confidence level is too low.

[0060] For example, if the confidence level of a candidate positioning policy is lower than the current confidence level, the system will prohibit replacing the current positioning policy with the candidate policy. Instead, it will continue positioning using the current policy, while simultaneously displaying a pop-up window on the central control screen indicating that the confidence level is too low, effectively improving vehicle interactivity.

[0061] In some other embodiments, after selectively calculating the confidence level of a candidate positioning policy, if the confidence level of the candidate positioning policy is lower than the current confidence level, it can be determined whether the current confidence level is lower than a preset unreliable confidence level. If the current confidence level is lower than the preset unreliable confidence level, a positioning failure notification is displayed.

[0062] For example, the embodiment of the present invention can determine whether the current reliability is lower than the unreliable reliability, and if the current reliability is lower than the unreliable reliability, it can improve the user's driving safety by displaying a pop-up window on the central control screen indicating that positioning is impossible.

[0063] According to the positioning method for autonomous vehicle driving proposed by the embodiment of the present invention, the current confidence level of the current positioning policy is calculated by obtaining the GDP value of the vehicle's current positioning policy. If the current confidence level is lower than the confidence level, the current positioning policy is replaced with a candidate positioning policy, and the vehicle's current positioning can be obtained based on the fusion of the vehicle's GNSS data, lidar data, and inertial measurement data. This effectively improves the accuracy of vehicle positioning when the GNSS signal is relatively weak, further enhances vehicle safety, and simultaneously satisfies the user's driving experience.

[0064] Next, a positioning device for autonomous vehicle driving proposed by the embodiment of the present invention will be described with reference to the drawings.

[0065] Figure 5 is a schematic block diagram of a positioning device in an automated vehicle driving system according to an embodiment of the present invention.

[0066] As shown in Figure 5, the positioning device 10 for the autonomous driving of the vehicle includes an acquisition module 100, a calculation module 200, and a processing module 300.

[0067] Of these, the acquisition module 100 is used to acquire the GDPR value of the vehicle's current positioning policy, the GDPR value is used to indicate the positioning accuracy of the vehicle's GNSS data, and the current positioning policy is the GNSS positioning policy.

[0068] The calculation module 200 is used to calculate the current confidence level of the current positioning policy based on the GDP values.

[0069] The processing module 300 determines that the current positioning policy is invalid if the current confidence level is lower than the confidence level, replaces the current positioning policy with a candidate positioning policy, and uses the candidate positioning policy to determine the vehicle's current positioning. The candidate positioning policy is a positioning policy that obtains the vehicle's current positioning based on the fusion of the vehicle's Global Navigation Satellite System (GNSS) data, LiDAR data, and inertial measurement data.

[0070] Selectively, in one embodiment of the present invention, the processing module 300 determines the initial position and attitude information of the vehicle in the map coordinate system based on the GNSS data and the inertial measurement data, and based on the initial position and attitude information and the LiDAR data, a point cloud scene earth A diagram is generated and used to determine the current position of the vehicle based on the matching result between the point cloud data in the LiDAR data and the point cloud scene map.

[0071] Selectively, in one embodiment of the present invention, the processing module 300 is used to determine global positioning information in a reference coordinate system based on the fusion of the GNSS data and the inertial measurement data by an extended Kalman filtering method, and to determine the transformation relationship between the reference coordinate system and the map coordinate system, and the initial position and attitude information of the vehicle in the map coordinate system based on the global positioning information.

[0072] Selectively, in one embodiment of the present invention, the processing module 300 is used to construct a point cloud map based on at least one keyframe corresponding to the initial position and orientation information of the LiDAR data, extract keyframe data from the LiDAR data, extract at least one point surface feature from the keyframe data, perform position and orientation matching for each point surface feature to obtain position and orientation information of the keyframes, insert point cloud data into the point cloud map based on the position and orientation information of the keyframes, and generate a point cloud scene map.

[0073] Selectively, in one embodiment of the present invention, the acquisition module is further used to acquire the residual Euclidean distance between point cloud data and corresponding points in the point cloud scene map before replacing the current positioning policy with a candidate positioning policy. The calculation module is further used to calculate the confidence level of the candidate positioning policy based on the residual Euclidean distance before replacing the current positioning policy with a candidate positioning policy. The replacement module is further used to perform the operation of replacing the current positioning policy with a candidate positioning policy if the confidence level of the candidate positioning policy is equal to or greater than the current confidence level, before replacing the current positioning policy with a candidate positioning policy.

[0074] Selectively, in one embodiment of the present application, the apparatus 10 of the embodiment of the present application further includes a control module. The control module is used to calculate the confidence level of a candidate positioning policy, and if the confidence level of the candidate positioning policy is lower than the current confidence level, to perform positioning using the current positioning policy as is, while simultaneously indicating that the confidence level is too low.

[0075] Selectively, in one embodiment of the present application, the apparatus 10 of the embodiment further includes a decision module, which is used to determine whether the confidence level of a candidate positioning policy is lower than the current confidence level, and whether the current confidence level is lower than the unreliable confidence level. A processing module is further used to indicate that positioning is impossible if the current confidence level is lower than the unreliable confidence level.

[0076] The interpretation and explanation of the aforementioned embodiment of the positioning method in the autonomous driving of a vehicle also applies to the positioning device in the autonomous driving of the vehicle in that embodiment, and therefore, the explanation is omitted here.

[0077] According to the positioning device for autonomous driving of a vehicle proposed by the embodiment of the present invention, the current confidence level of the current positioning policy is calculated by obtaining the GDP value of the vehicle's current positioning policy. If the current confidence level is lower than the confidence level, the current positioning policy is replaced with a candidate positioning policy, and the vehicle's current positioning can be obtained based on the fusion of the vehicle's GNSS data, lidar data, and inertial measurement data. This effectively improves the accuracy of vehicle positioning when the GNSS signal is relatively weak, thereby improving vehicle safety while also enhancing the user's driving experience.

[0078] Figure 6 is a schematic diagram of the structure of a vehicle provided according to an embodiment of the present application. This vehicle is, The system may also include memory 601, a processor 602, and a computer program stored in memory 601 and executable by the processor 602.

[0079] The processor 602 implements the positioning method for autonomous driving of a vehicle provided in the above embodiment when executing the program.

[0080] Furthermore, the vehicle, A communication interface 603 for communication between memory 601 and processor 602, The system further includes memory 601 for storing computer programs that can be executed on processor 602.

[0081] The memory 601 may include high-speed RAM memory and may further include at least one non-volatile memory such as magnetic disk memory.

[0082] If the memory 601, processor 602, and communication interface 603 are implemented independently, the communication interface 603, memory 601, and processor 602 can be interconnected via a bus and communicate with each other completely. The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of simplicity, it is shown by only one thick line in Figure 6, but this does not mean that only one bus or one type of bus exists.

[0083] If, selectively and in a concrete implementation, the memory 601, processor 602, and communication interface 603 are integrated onto a single chip, the memory 601, processor 602, and communication interface 603 can communicate with each other through an internal interface.

[0084] The processor 602 may be a single central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits arranged to carry out the embodiments of the present application.

[0085] This embodiment further provides a computer-readable storage medium in which a computer program is stored, and which realizes the positioning method for the automated driving of the vehicle described above when the computer program is executed by a processor.

[0086] The embodiments of the present invention further provide a computer program product that enables a vehicle to implement the above-described positioning method for autonomous driving of a vehicle when executed in the vehicle.

[0087] In this specification, any reference to terms such as “one embodiment,” “several embodiments,” “example,” “specific example,” or “several examples” means that the specific features, structures, materials, or properties described in combination with such embodiment or example are included in at least one embodiment or example of this application. In this specification, the descriptive expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or properties described may be combined in an appropriate manner in any one or N embodiments or examples. Moreover, a person skilled in the art can combine different embodiments or examples and features of different embodiments or examples described herein without contradiction.

[0088] Furthermore, the terms “first” and “second” are merely for descriptive purposes and should not be understood as indicating or implying relative importance, or implicitly indicating the quantity of the technical features being referred to. Accordingly, features designated as “first” or “second” may explicitly or implicitly include at least one such feature. In this description, “N” means at least two, for example, two, three, etc., unless otherwise clearly and specifically defined.

[0089] A flowchart or any description of a process or method described herein in any other way can be understood to represent a module, segment, or portion containing code for one or N executable instructions to implement steps for customizing a logical function or process, and the scope of preferred embodiments of the present application includes other implementations, which may not follow the order shown or discussed, and which include performing functions based on such functions in essentially simultaneous or reverse order, as should be understood by those skilled in the art to which the embodiments of the present application belong.

[0090] The logic and / or steps shown in the flowchart or otherwise described herein can be considered, for example, an ordered list of executable instructions for realizing a logical function, which can be specifically implemented on any computer-readable medium for use by an instruction execution system, device or apparatus (e.g., a computer-based system, a system including a processor, or other system capable of obtaining and executing instructions from an instruction execution system, device or apparatus), or for use in combination with such instruction execution systems, devices or apparatus. In this specification, “computer-readable medium” may be any device capable of containing, storing, communicating, propagating, or transmitting programs for use by an instruction execution system, device or apparatus, or for use in combination with such instruction execution systems, devices or apparatus. More specific examples of computer-readable mediums (a non-exclusive list) include electrical connections with one or N wires (electronic devices), portable computer disk boxes (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and compact disk read-only memory (CDROM). Furthermore, the computer-readable medium may be paper or other suitable medium on which the above program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, and then editing, interpreting, or processing it in any other suitable way as necessary, and then stored in computer memory.

[0091] It should be understood that each part of this application can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in memory and executed by an appropriate instruction execution system. For example, if implemented in hardware, it can be implemented in any or a combination thereof of technologies known in the art, such as discrete logic circuits having logic gate circuits for implementing logic functions for data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), and field-programmable gate arrays (FPGAs), as with other implementations.

[0092] Those skilled in the art will understand that all or some of the steps included in the method of the above embodiment can be completed by instructing the relevant hardware by a program, the program of which can be stored on a computer-readable storage medium, and when the program is executed, it includes one or a combination of the steps of the embodiment of the method.

[0093] Furthermore, each functional unit in each embodiment of the present application may be integrated into a single processing module, each unit may exist individually in physical form, or two or more units may be integrated into a single module. The integrated module may be implemented in hardware form or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may be stored in a single computer-readable storage medium.

[0094] The storage medium mentioned above may be a read-on memory, magnetic disk, or optical disk, etc. Although embodiments of the present application have been described above, these embodiments are illustrative and should not be understood as limiting the present application, and those skilled in the art will understand that the above embodiments can be modified, altered, substituted, or transformed within the scope of the present application.

Claims

1. A positioning method for autonomous driving of a vehicle, wherein the method is A step of obtaining a GDPR (Geometric Degradation of Positioning) value for the vehicle's current positioning policy, wherein the GDPR value is used to indicate the positioning accuracy of the vehicle's Global Navigation Satellite System (GNSS) data, and the current positioning policy is a GNSS positioning policy. A step of calculating the current confidence level of the current positioning policy based on the GDP figures, The steps include: determining that the current positioning policy is invalid if the current confidence level is lower than a reliable confidence level, replacing the current positioning policy with a candidate positioning policy, and determining the current position of the vehicle based on the candidate positioning policy, wherein the candidate positioning policy is a positioning policy that obtains the current position of the vehicle based on the fusion of the vehicle's GNSS data, lidar data, and inertial measurement data, A method for determining position in autonomous vehicle driving.

2. Obtaining the current position of the vehicle based on the fusion of the aforementioned GNSS data, lidar data, and inertial measurement data of the vehicle is possible. Based on the GNSS data and the inertial measurement data, the initial position and attitude information of the vehicle in the map coordinate system is determined. Based on the initial position and orientation information and the lidar data, a point cloud scene diagram is generated. The method is characterized by including determining the current positioning of the vehicle based on the matching result between the point cloud data in the LiDAR data and the point cloud scene map. The method according to claim 1.

3. Based on the aforementioned GNSS data and inertial measurement data, determining the initial position and attitude information of the vehicle in the map coordinate system is: The extended Kalman filtering method determines global positioning information in the reference coordinate system based on the GNSS data and the inertial measurement data. The method is characterized by including determining the initial position and orientation information of the vehicle in the map coordinate system based on the transformation relationship between the reference coordinate system and the map coordinate system, and the global positioning information. The method according to claim 2.

4. Based on the aforementioned initial position and orientation information and the lidar data, generating a point cloud scene diagram is possible. A point cloud map is constructed based on at least one keyframe from the aforementioned lidar data that corresponds to the initial position and orientation information. Extracting keyframe data from the aforementioned RIDGID data, Extracting at least one point-plane feature from the aforementioned keyframe data, performing position-orientation matching on each point-plane feature, and obtaining position-orientation information for the keyframes, The method is characterized by including the following: inserting point cloud data from the lidar data into the point cloud map based on the position and orientation information of the keyframes, and generating a point cloud scene map. The method according to claim 2.

5. Before replacing the current positioning policy with the candidate positioning policy, the method: The steps include obtaining the residual Euclidean distance between the point cloud data and the corresponding point in the point cloud scene map, A step of calculating the confidence level of the candidate positioning policy based on the residual of the Euclidean distance, The method further includes the step of replacing the current positioning policy with the candidate positioning policy if the confidence level of the candidate positioning policy is equal to or greater than the current confidence level. The method described in claim 2

6. After calculating the confidence level of the candidate positioning policy based on the residual of the Euclidean distance described above, the method proceeds as follows: If the confidence level of the candidate positioning policy is lower than the current confidence level, the method further includes the step of performing positioning using the current positioning policy as is, while simultaneously indicating that the confidence level is too low. The method according to claim 5.

7. After calculating the confidence level of the candidate positioning policy based on the residual of the Euclidean distance described above, the method proceeds as follows: If the confidence level of the candidate positioning policy is lower than the current confidence level, the step is to determine whether the current confidence level is lower than the unreliable confidence level. The present invention further includes the step of indicating that positioning is impossible if the current level of confidence is lower than the unreliable level of confidence, The method according to claim 5.

8. A positioning device for autonomous driving of a vehicle, wherein the device is An acquisition module for obtaining a GDPR (Geometric Degradation Rate) value of a vehicle's current positioning policy, wherein the GDPR value is used to indicate the positioning accuracy of the vehicle's Global Navigation Satellite System (GNSS) data, and the current positioning policy is a GNSS positioning policy. A calculation module for calculating the current confidence level of the current positioning policy based on the GDP figures, A processing module for determining that the current positioning policy is invalid if the current confidence level is lower than a reliable confidence level, replacing the current positioning policy with a candidate positioning policy, and determining the current positioning of the vehicle based on the candidate positioning policy, wherein the candidate positioning policy is a positioning policy that obtains the current positioning of the vehicle based on the fusion of the vehicle's GNSS data, lidar data, and inertial measurement data, includes: A positioning device used in autonomous vehicle driving.

9. A vehicle comprising memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor executes the computer program in order to realize a positioning method for the automated driving of the vehicle according to any one of claims 1 to 7. vehicle.

10. A computer-readable storage medium in which a computer program is stored, characterized in that the computer program is executed by a processor in order to realize the positioning method for the automated driving of a vehicle described in any one of claims 1 to 7. Computer-readable storage medium.

Images