Combination positioning method and device based on graph optimization, electronic equipment and medium

Abstract

Embodiments of the present application relate to the field of integrated positioning, and in particular to an integrated positioning method and device based on graph optimization, electronic equipment and a medium. A specific embodiment of the method comprises: generating a positioning factor graph; obtaining target laser radar de-distortion information; obtaining radar coordinate information; determining target anchor point information; determining an initial heading angle and residual information; in response to determining that pre-acquired observation mode information satisfies a preset mode condition, performing coordinate conversion processing on RTK positioning information based on the positioning factor graph, the target anchor point information, the initial heading angle and the residual information to obtain target positioning information. This embodiment greatly improves the accuracy of integrated positioning.

Description

Technical Field

[0001] Embodiments of this application relate to the field of combined positioning, specifically to graph-optimized combined positioning methods, apparatuses, electronic devices, and media. Background Technology

[0002] With the rapid development of robots and unmanned systems, the requirements for positioning accuracy and robustness are increasing, and the scenarios they face are becoming more diversified. Global Navigation Satellite Systems (GNSS) provide users with high-precision location information. Real-time kinematic (RTK) is widely used due to its high accuracy and ease of operation; however, its accuracy and robustness decrease significantly when satellite signals are blocked for extended periods, especially in complex modern urban environments. Inertial Navigation Systems (INS) can achieve short-term high-precision positioning without environmental dependence, but as time progresses, errors accumulate due to noise from the Inertial Measurement Unit (IMU). Light Detection and Ranging (LiDAR) has advantages such as small cumulative errors and immunity to light, and can provide high-precision local pose for vehicles by scanning rich features in complex urban environments; however, it still faces drawbacks such as point cloud distortion during motion and poor accuracy in open environments.

[0003] In conclusion, single-sensor positioning is insufficient for high-precision positioning in complex urban environments. Therefore, integrating the advantages of different sensors to leverage their strengths and improve the accuracy and robustness of multi-source fusion positioning systems is a current research focus in the field of navigation and positioning. Summary of the Invention

[0004] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0005] Some embodiments of this application propose a graph-optimized combined localization method, apparatus, computer device, and computer-readable storage medium to solve one or more of the technical problems mentioned in the background section above.

[0006] In a first aspect, some embodiments of this application provide a graph-optimized combined positioning method, which includes: in response to determining that pre-acquired historical lidar observation information meets preset initialization conditions, generating a positioning factor map based on the historical lidar observation information, pre-acquired historical IMU observation information, and historical RTK positioning information; performing distortion correction processing on the pre-acquired lidar observation information based on the positioning factor map to obtain target lidar distortion correction information; performing coordinate transformation processing on the target lidar distortion correction information based on the positioning factor map to obtain radar coordinate information; determining target anchor point information based on the target lidar distortion correction information and pre-acquired RTK positioning information; determining initial heading angle and residual information based on the positioning factor map, the radar coordinate information, and the RTK positioning information; and in response to determining that pre-acquired observation mode information meets preset mode conditions, performing coordinate transformation processing on the RTK positioning information based on the positioning factor map, the target anchor point information, the initial heading angle, and the residual information to obtain target positioning information.

[0007] Secondly, some embodiments of this disclosure provide a graph-optimized combined positioning device, comprising: a generation unit configured to generate a positioning factor map based on the historical lidar observation information, the pre-acquired historical IMU observation information, and the pre-acquired historical RTK positioning information in response to determining that the pre-acquired historical lidar observation information meets preset initialization conditions; a distortion correction unit configured to perform distortion correction processing on the pre-acquired lidar observation information based on the positioning factor map to obtain target lidar distortion correction information; and a first coordinate transformation unit configured to perform a first coordinate transformation on the target lidar distortion correction information based on the positioning factor map. The system performs coordinate transformation to obtain radar coordinate information; a first determining unit is configured to determine target anchor point information based on the target lidar distortion correction information and pre-acquired RTK positioning information; a second determining unit is configured to determine initial heading angle and residual information based on the positioning factor map, radar coordinate information, and RTK positioning information; and a second coordinate transformation unit is configured to, in response to determining that the pre-acquired observation mode information meets preset mode conditions, perform coordinate transformation on the RTK positioning information based on the positioning factor map, target anchor point information, initial heading angle, and residual information to obtain target positioning information.

[0008] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0009] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0010] The above embodiments of this application have the following beneficial effects: The graph-optimized integrated positioning method of some embodiments of this application addresses the problem that RTK cannot achieve high-precision and robust positioning in complex urban environments. This paper proposes an integrated positioning algorithm. It constructs LiDAR odometry factors, IMU pre-integration factors, and RTK factors, and optimizes them using factor graphs. During the algorithm initialization phase, anchor points are obtained using interpolation, and the initial heading angle is optimized using factor graph optimization to achieve coordinate system transformation and unification. Simultaneously, an RTK filtering strategy is used to filter out poor-quality RTK information, and weights are dynamically allocated. A forced RTK addition strategy is added to prevent excessive cumulative errors caused by prolonged RTK non-fusion. Centimeter-level accuracy can be achieved in open environments, decimeter-level accuracy in complex urban environments, and stable high-precision positioning results can be provided even when RTK signals are interrupted. Future work can explore achieving a tight combination of RTK and LIO, and adding new sensors, such as visual sensors, to further improve the positioning performance of integrated navigation in complex environments. Attached Figure Description

[0011] The above and other features, advantages, and aspects of the embodiments of this application will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0012] Figure 1 This is a flowchart of some embodiments of the graph optimization-based combined localization method according to this application;

[0013] Figure 2 This is a fusion accuracy map in an open environment based on some embodiments of the graph optimization-based combined localization method of this application;

[0014] Figure 3 This is a comparison table of root mean square error values ​​in open environments based on some embodiments of the graph optimization-based combined positioning method of this application.

[0015] Figure 4 This is an error fluctuation diagram in an open environment based on some embodiments of the graph optimization-based combined positioning method according to this application;

[0016] Figure 5 This is a vehicle trajectory map in an open environment according to some embodiments of the graph optimization-based combined localization method of this application;

[0017] Figure 6 This is an overall accuracy map of driving from an open area into a complex urban environment, based on some embodiments of the graph optimization-based combined positioning method of this application.

[0018] Figure 7 This is an accuracy map of the preceding open environment during the process of driving from an open area into a complex urban environment, according to some embodiments of the graph optimization-based combined positioning method of this application.

[0019] Figure 8 These are vehicle trajectory maps of vehicles entering complex urban environments from open spaces, according to some embodiments of the graph optimization-based combined localization method of this application.

[0020] Figure 9 This is a comparison table of root mean square error values ​​for driving from an open area into a complex urban environment, based on some embodiments of the graph optimization-based combined positioning method of this application.

[0021] Figure 10 This is an overall accuracy map under signal interruption in a complex urban environment according to some embodiments of the graph optimization-based combined positioning method of this application;

[0022] Figure 11 This is a comparison table of root mean square error values ​​under signal interruption in complex urban environments according to some embodiments of the graph optimization-based combined localization method of this application;

[0023] Figure 12 These are schematic diagrams of some embodiments of the graph-optimized combined positioning device according to this application;

[0024] Figure 13 This is a schematic diagram of the structure of a computer device suitable for implementing some embodiments of this application. Detailed Implementation

[0025] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0026] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0027] It should be noted that the concepts of "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0028] It should be noted that the terms "a" and "a plurality of" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0029] The names of the messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0030] The present application will now be described in detail with reference to the accompanying drawings and embodiments.

[0031] Figure 1 A flowchart 100 of some embodiments of the graph-optimized combinatorial localization method according to this application is shown. The graph-optimized combinatorial localization method includes the following steps:

[0032] Step 101: In response to determining that the pre-acquired historical lidar observation information meets the preset initialization conditions, a positioning factor map is generated based on the historical lidar observation information, the pre-acquired historical IMU observation information, and the historical RTK positioning information.

[0033] In some embodiments, the execution entity of the graph-optimized combined localization method can, in response to determining that pre-acquired historical lidar observation information meets preset initialization conditions, generate a localization factor graph based on the aforementioned historical lidar observation information, pre-acquired historical IMU observation information, and historical RTK localization information. This can be achieved through wired or wireless connections, acquiring historical lidar observation information from lidar sensors, historical IMU observation information from IMU sensors, and historical RTK localization information from GNSS. The aforementioned historical lidar observation information can characterize the laser data collected by the lidar within the historical time period. This historical lidar observation information may include a lidar observation keyframe sequence. The lidar observation keyframes in the aforementioned lidar observation keyframe sequence may include observation timestamps. The lidar observation keyframes in the aforementioned lidar observation keyframe sequence can be keyframes selected from the laser data collected by the lidar within the historical time period using a keyframe strategy. The observation timestamps can characterize the time at which the corresponding lidar observation keyframe was acquired. The aforementioned historical IMU observation information can characterize the IMU data collected by the IMU within the historical time period. The aforementioned historical IMU observation information may include: an IMU observation information sequence. The IMU observation information in the aforementioned IMU observation information sequence may include: IMU acquisition time, angular velocity, and acceleration. The aforementioned IMU acquisition time can represent the time when the corresponding IMU observation information was acquired. The aforementioned historical RTK positioning information can represent RTK data acquired by GNSS within a historical time period. The aforementioned angular velocity can represent the angular velocity acquired by the IMU. The aforementioned acceleration can represent the acceleration acquired by the IMU.

[0034] In practice, firstly, the difference information between every two adjacent keyframes in the keyframe sequence of historical lidar observations is determined, resulting in a difference information set. Here, a preset radar difference algorithm can be used to determine the difference information between every two adjacent keyframes in the keyframe sequence of historical lidar observations. This difference information may include radar displacement values ​​and radar angle change values. Then, the preset initialization condition can be that the difference information set contains difference information where the radar displacement value is greater than a preset displacement value or the radar angle change value is greater than a preset angle change value.

[0035] As an example, the preset radar difference algorithm mentioned above can be a point cloud registration algorithm. The preset displacement value mentioned above can be 0.05 meters. The preset angle change value mentioned above can be 0.2 rad (revolution).

[0036] It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future wireless connection methods.

[0037] In some optional implementations of certain embodiments, the execution entity generates a positioning factor map based on the aforementioned historical lidar observation information and pre-acquired historical IMU observation information and historical RTK positioning information, which may include the following steps:

[0038] The first step is to perform distortion correction on the historical IMU observation information to obtain IMU pre-integration factor information and lidar observation information.

[0039] The second step is to construct radar odometer factor information based on the aforementioned lidar observation information.

[0040] The third step is to construct RTK factor information based on the aforementioned historical RTK positioning information and lidar observation information.

[0041] The fourth step involves fusing the aforementioned radar odometer factor information, IMU pre-integration factor information, and RTK factor information to obtain a positioning factor map. Specifically, fusing these factors to obtain the positioning factor map can be achieved by defining the radar odometer factor information, IMU pre-integration factor information, and RTK factor information as the radar odometer factor information, IMU pre-integration factor information, and RTK factor information included in the positioning factor map.

[0042] Here, in terms of weight selection for fusion, data is filtered and weighted based on the number of observable satellites and the PDOP (Position Dilution of Precision) value in the RTK data. At the same time, the function of forcibly adding RTK data is added to ensure that the positioning will not diverge when stationary for a long time in complex environments.

[0043] In some optional implementations of certain embodiments, the execution entity performs distortion correction processing on the historical IMU observation information based on the historical IMU observation information to obtain IMU pre-integration factor information and lidar observation information, which may include the following steps:

[0044] The first step is to perform the following distortion correction sub-step for each lidar observation keyframe in the lidar observation keyframe sequence included in the above historical lidar observation information:

[0045] The first sub-step is to determine the observation timestamps included in the aforementioned lidar observation keyframes as the first timestamp.

[0046] The second sub-step involves determining the observation timestamp included in the next lidar observation keyframe within the aforementioned lidar observation keyframe sequence as the second timestamp.

[0047] The third sub-step involves constructing the IMU pre-integration factor corresponding to the aforementioned historical IMU observation information, based on the first and second timestamps mentioned above.

[0048] The fourth sub-step involves generating prior radar pose information based on the aforementioned IMU pre-integration factor. This prior radar pose information can be generated using a preset pose generation algorithm based on the IMU pre-integration factor.

[0049] As an example, the preset pose generation algorithm mentioned above may be, but is not limited to, at least one of the following: ICP (Iterative Closest Point) algorithm or LIO-SAM (Tightly-coupled Lidar Inertial Odometry via Smoothing and Mapping) algorithm.

[0050] The fifth sub-step involves performing distortion correction processing on the aforementioned lidar observation keyframes based on the prior radar pose information, resulting in distorted lidar frames. Specifically, based on the prior radar pose information, a preset distortion correction algorithm can be used to perform distortion correction processing on the aforementioned lidar observation keyframes to obtain distorted lidar frames. These distorted lidar frames include a lidar point set. The lidar points in the lidar point set can represent a point in the lidar point cloud.

[0051] As an example, the aforementioned preset distortion correction algorithm can be, but is not limited to, at least one of the following: linear interpolation or the LIO-SAM algorithm.

[0052] The second step is to determine the generated LiDAR distortion-corrected frames as a LiDAR distortion-corrected frame sequence.

[0053] The third step is to determine the generated IMU pre-integration factors as IMU pre-integration factor information.

[0054] The fourth step is to determine the above-mentioned LiDAR distortion-corrected frame sequence as LiDAR observation information.

[0055] In some optional implementations of certain embodiments, the execution entity constructs the IMU pre-integration factor corresponding to the historical IMU observation information based on the first timestamp and the second timestamp, which may include the following steps:

[0056] The first step is to determine the difference between the first timestamp and the second timestamp as the predicted duration.

[0057] The second step involves, based on the aforementioned prediction duration, performing prediction processing on the IMU observation information corresponding to the first timestamp within the IMU observation information sequence included in the aforementioned historical IMU observation information, to obtain the IMU prediction information. Specifically, based on the aforementioned prediction duration, the IMU prediction information can be obtained by performing prediction processing on the IMU observation information corresponding to the first timestamp within the IMU observation information sequence included in the aforementioned historical IMU observation information using the following formula:

[0058]

[0059] .

[0060] in, This indicates the first timestamp mentioned above. This indicates the angular velocity included in the IMU observation information sequence that corresponds to the first timestamp mentioned above, within the aforementioned historical IMU observation information. This represents the true value of the angular velocity corresponding to the IMU observation information corresponding to the first timestamp mentioned above, within the IMU observation information sequence included in the aforementioned historical IMU observation information. This represents the zero bias value of the IMU's angular velocity. This represents the white noise value of the IMU's angular velocity. This indicates the acceleration included in the IMU observation information sequence that corresponds to the first timestamp mentioned above, within the aforementioned historical IMU observation information. This represents the true value of the acceleration corresponding to the IMU observation information corresponding to the first timestamp mentioned above, within the IMU observation information sequence included in the aforementioned historical IMU observation information. This represents the zero bias value of the IMU's acceleration. This represents the white noise value indicating the acceleration of the IMU. express The rotation matrix from the W coordinate system to the b coordinate system at time t. This represents the constant gravity vector in the W-frame coordinate system.

[0061] This indicates the predicted duration. This indicates the speed included in the above IMU prediction information. This indicates the velocity of the IMU corresponding to the IMU observation information corresponding to the first timestamp mentioned above, within the IMU observation information sequence included in the aforementioned historical IMU observation information. . This indicates the locations included in the above IMU prediction information. This indicates the location of the IMU corresponding to the IMU observation information corresponding to the first timestamp mentioned above, within the IMU observation information sequence included in the aforementioned historical IMU observation information. This indicates the rotation matrix included in the above IMU prediction information.

[0062] Here, the W-coordinate system mentioned above can be the world coordinate system. The world coordinate system is a common coordinate system in SLAM (Simultaneous Localization and Mapping) algorithms. Its coordinate axes are orthogonal to each other and conform to the right-hand rule. The origin and the orientation of the coordinate axes can be arbitrarily specified according to different requirements. In this paper, the origin of the world coordinate system is chosen at the position of the first radar frame, set as (0, 0, 0). At this time, the z-axis is determined according to the IMU gravity direction, the initial direction of the vehicle's movement is the x-axis, and the y-axis is determined according to the right-hand rule. This constitutes a front-left-top coordinate system.

[0063] The aforementioned b-coordinate system can be the carrier coordinate system (b-coordinate system). The carrier coordinate system takes the center of the inertial sensor as its origin, with the x-axis perpendicular to the carrier's forward direction and pointing to the right, the y-axis pointing to the carrier's forward direction, and the z-axis perpendicular to the carrier's motion plane and pointing upward along the carrier's vertical axis, forming a right-handed system, namely the right-front-up carrier coordinate system.

[0064] The third step involves pre-integrating the historical IMU observation information and the IMU prediction information to obtain the IMU pre-integration factor. This pre-integration factor includes the velocity, position, and rotation matrices. The pre-integration process of the historical IMU observation information and the IMU prediction information to obtain the IMU pre-integration factor can be achieved using the following formula:

[0065] .

[0066] in, This indicates the second timestamp mentioned above. This indicates the speed included in the aforementioned pre-integral factor. This indicates the velocity of the IMU corresponding to the IMU observation information corresponding to the second timestamp mentioned above, within the IMU observation information sequence included in the aforementioned historical IMU observation information. This indicates the predicted duration. This indicates the positions included in the aforementioned pre-integral factors. This indicates the location of the IMU corresponding to the IMU observation information corresponding to the second timestamp in the IMU observation information sequence included in the above historical IMU observation information. This represents the rotation matrix included in the aforementioned pre-integral factors. . express The rotation matrix from the W coordinate system to the b coordinate system at time t.

[0067] In some optional implementations of certain embodiments, the execution entity constructs radar odometry factor information based on the aforementioned lidar observation information, which may include the following steps:

[0068] The first step, for each lidar distortion-corrected frame in the lidar distortion-corrected frame sequence included in the above lidar observation information, is to perform the following determination sub-step:

[0069] The first sub-step involves determining the curvature feature information corresponding to each lidar point in the lidar point set included in the lidar distortion correction frame, thereby obtaining a curvature feature information set. Specifically, determining the curvature feature information corresponding to each lidar point in the lidar point set included in the lidar distortion correction frame can be achieved by first determining the curvature value of the lidar point. Then, in response to determining that the curvature value of the lidar point is greater than a preset curvature value, a first preset feature information is determined as curvature feature information. In response to determining that the curvature value of the lidar point is less than or equal to the preset curvature value, a second preset feature information is determined as curvature feature information.

[0070] As an example, the aforementioned preset curvature values ​​can be relative. For instance, all points in the current frame are scanned and their curvature values ​​are calculated. Then, they are sorted from largest to smallest, with the top 10% of points with the highest curvature set as edge points and the bottom 20% of points with the lowest curvature set as planar points. The aforementioned first preset feature information can be information representing "edge features." The aforementioned second preset feature information can be information representing "planar features."

[0071] Specifically, the curvature value of the laser point corresponding to the above-mentioned lidar point can be determined using the following formula:

[0072] .

[0073] in, This represents the curvature value of the laser point. Indicates scan line One point in it. This indicates the aforementioned lidar points. Point The set of surrounding points. express One point in it. Point Coordinates in the lidar system. Point Coordinates in the lidar system.

[0074] The second sub-step involves classifying the aforementioned LiDAR point set based on the curvature feature information set to obtain a LiDAR planar point set and a LiDAR edge point set. Specifically, the classification of the LiDAR point set based on the curvature feature information set to obtain the LiDAR planar point set and the LiDAR edge point set can be achieved by: determining the LiDAR points in the aforementioned LiDAR point set that correspond to each curvature feature information representing a first preset feature information in the aforementioned curvature feature information set as the LiDAR edge point set; and determining the LiDAR points in the aforementioned LiDAR point set that correspond to each curvature feature information representing a second preset feature information in the aforementioned curvature feature information set as the LiDAR planar point set.

[0075] The third sub-step involves downsampling the aforementioned lidar planar point set and lidar edge point set to obtain the lidar sampling planar point set and lidar sampling edge point set, respectively. This downsampling can be performed using a preset downsampling algorithm.

[0076] As an example, the preset downsampling algorithm mentioned above may be, but is not limited to, at least one of the following: voxel grid downsampling algorithm, uniform downsampling algorithm, curvature downsampling algorithm, or random downsampling algorithm.

[0077] The fourth sub-step involves determining the voxel map corresponding to the next LiDAR distortion frame in the aforementioned LiDAR distortion frame sequence. This determination can be achieved using a pre-defined conversion algorithm.

[0078] As an example, the aforementioned preset conversion algorithm may be, but is not limited to, at least one of the following: a binary occupancy grid algorithm or a probabilistic occupancy grid algorithm.

[0079] The fifth sub-step involves determining the planar feature distance value corresponding to each lidar sampling plane point in the lidar sampling plane point set, based on the aforementioned voxel map, thus obtaining a set of planar feature distance values. Specifically, determining the planar feature distance value corresponding to each lidar sampling plane point in the aforementioned lidar sampling plane point set, based on the aforementioned voxel map, can be achieved using the following formula:

[0080] .

[0081] in, Indicates the serial number. The first point in the lidar sampling plane point set represents the... The planar feature distance value corresponding to each sampling plane point of the lidar. , , and Indicates the formation of the first voxel map The coordinates of the points on the plane corresponding to the sampling points of each lidar.

[0082] The sixth sub-step involves determining the edge feature distance value corresponding to each lidar sampling edge point in the lidar sampling edge point set, based on the aforementioned voxel map, thus obtaining a set of edge feature distance values. The determination of the edge feature distance value corresponding to each lidar sampling edge point in the aforementioned lidar sampling edge point set, based on the aforementioned voxel map, can be achieved using the following formula:

[0083] .

[0084] in, Indicates the first point in the lidar sampling edge point set The edge feature distance value corresponding to each edge point sampled by the LiDAR. , and Indicates the formation of the first voxel map The coordinates of the points on the edge line corresponding to the edge points sampled by the lidar.

[0085] The seventh sub-step involves determining the pose transformation matrix based on the aforementioned planar feature distance value set and the aforementioned edge feature distance value set. Specifically, the pose transformation matrix can be determined using a preset pose determination algorithm based on the aforementioned planar feature distance value set and the aforementioned edge feature distance value set.

[0086] As an example, the preset pose transformation algorithm mentioned above can be the Gauss-Newton method.

[0087] The eighth sub-step involves determining the radar odometry factor based on the aforementioned pose transformation matrix. The radar odometry factor can be determined using the following formula:

[0088] .

[0089] in, Indicates the serial number. This indicates that the above-mentioned lidar observation information includes the lidar distortion correction frame sequence of the [number]th [frame]. The radar odometry factor corresponding to each LiDAR distortion-corrected frame. This indicates that the above-mentioned lidar observation information includes the lidar distortion correction frame sequence of the [number]th [frame]. The pose transformation matrix corresponding to each LiDAR distortion-free frame. This indicates that the above-mentioned lidar observation information includes the lidar distortion correction frame sequence of the [number]th [frame]. The pose transformation matrix corresponding to the next LiDAR distortion-corrected frame of the previous LiDAR distortion-corrected frame.

[0090] The second step is to determine the generated radar odometry factors as radar odometry factor information.

[0091] In some optional implementations of certain embodiments, the execution entity constructs RTK factor information based on the aforementioned historical RTK positioning information and the aforementioned lidar observation information, which may include the following steps:

[0092] The first step is to determine the historical anchor point information based on the aforementioned lidar observation information and historical RTK positioning information. This determination can be achieved by setting the anchor point as the origin of the world coordinate system (i.e., the origins of the world coordinate system and the navigation coordinate system coincide), and setting the coordinates of the RTK in the ECEF (geocentric-ground-fixed) system at the moment corresponding to the first lidar distortion-corrected frame in the lidar distortion-corrected frame sequence included in the lidar observation information as the historical anchor point information. Specifically, this historical anchor point information can be obtained by interpolating the RTK positioning data before and after the first lidar frame.

[0093] The ECEF system (geocentric-fixed coordinate system) mentioned above can be: with the Earth's center as the origin, the x-y coordinate plane is parallel to the Earth's equatorial plane, the x-axis points from the Earth's center to the prime meridian, the z-axis is perpendicular to the Earth's equatorial plane and points to the geographic North Pole, and finally the y-axis is taken to make the ECEF coordinate system a right-handed system.

[0094] The second step is to determine the initial heading angle. This can be achieved by defining the horizontal angle between the W and n frames of reference.

[0095] Here, the n-system represents the navigation coordinate system: a navigation coordinate system is introduced to connect the world coordinate system and the Earth-centered Earth-fixed system. The x-axis points east, the y-axis points north, and the z-axis forms a right-handed coordinate system. This is called the East-North-Sky coordinate system. As a transitional system, if the coordinates of a point in the ECEF coordinate system are known, a unique navigation coordinate system can be determined based on this point, with the origin located at that point. In this study, this point is selected as the measurement center of the first radar frame in the ECEF system.

[0096] The third step involves constructing a first rotation matrix and a second rotation matrix based on the aforementioned historical lidar observation information and initial heading angle. Specifically, the first and second rotation matrices can be constructed using the following formulas:

[0097]

[0098] .

[0099] in, This represents the first rotation matrix mentioned above. It can characterize the rotation matrix from the W system to the n system. This indicates the initial heading angle. This represents the second rotation matrix mentioned above. It can characterize the rotation matrix from the n-system to the ECEF system. and These represent the longitude and latitude of the lidar unit that collected the aforementioned historical lidar observation information at the initial moment. The initial moment can be a time when no historical lidar observation information was collected.

[0100] Fourth, based on the aforementioned historical anchor point information, the first rotation matrix, and the second rotation matrix, construct the coordinate transformation formula. Specifically, based on the aforementioned historical anchor point information, the first rotation matrix, and the second rotation matrix, the following coordinate transformation formula can be constructed:

[0101] .

[0102] in, This indicates the location of the aforementioned GNSS in the ECEF system. This indicates the location of the aforementioned GNSS in the W system. This indicates the aforementioned historical anchor point information.

[0103] Here, since the GNSS measurement reference is the receiver phase center, while the reference reference in this application is the LiDAR measurement center, the two are spatially asynchronous, requiring the use of the GNSS center in the b-series. Relative to LiDAR center The vector (also known as the lever arm) Make corrections: .

[0104] The fifth step involves fusing the first rotation matrix, the second rotation matrix, and the coordinate transformation formula to obtain the RTK factor information. Specifically, fusing the first rotation matrix, the second rotation matrix, and the coordinate transformation formula to obtain the RTK factor information can be achieved by defining the first rotation matrix, the second rotation matrix, and the coordinate transformation formula as the first rotation matrix, the second rotation matrix, and the coordinate transformation formula included in the RTK factor information.

[0105] Optionally, during the initialization phase, the initial heading angle can be optimized 10 times to obtain a reliable initial value before the RTK factor can be added. At this point, the initialization is complete, and the initial heading angle will be continuously optimized in subsequent optimizations.

[0106] Step 102: Based on the positioning factor map, the pre-acquired lidar observation information is subjected to distortion correction processing to obtain the target lidar distortion correction information.

[0107] In some embodiments, the execution entity can perform distortion correction processing on the pre-acquired lidar observation information based on the positioning factor map to obtain the target lidar distortion correction information. The specific implementation method for generating the target lidar distortion correction information and its resulting technical effects can be found in step 101 of the above embodiments, and will not be repeated here.

[0108] Step 103: Based on the positioning factor map, perform coordinate transformation on the distortion-free information of the target lidar to obtain the radar coordinate information.

[0109] In some embodiments, the execution entity can perform coordinate transformation processing on the distortion-corrected information of the target lidar based on the positioning factor map to obtain radar coordinate information. Specifically, the IMU pre-integration factor information and radar odometry factor information included in the positioning factor map can be used to perform coordinate transformation processing on the distortion-corrected information of the target lidar using a preset radar coordinate transformation algorithm to obtain radar coordinate information.

[0110] As an example, the aforementioned preset radar coordinate algorithm can be the LIO-SAM algorithm.

[0111] Step 104: Determine the target anchor point information based on the target lidar distortion correction information and the pre-acquired RTK positioning information.

[0112] In some embodiments, the execution entity can determine the target anchor point information based on the target LiDAR distortion correction information and pre-acquired RTK positioning information. Specifically, determining the target anchor point information based on the target LiDAR distortion correction information and pre-acquired RTK positioning information can involve setting the anchor point as the origin of the world coordinate system (i.e., the origin of the world coordinate system coincides with the origin of the navigation coordinate system), and setting the RTK coordinates in the ECEF system at the time corresponding to the first frame of the LiDAR point cloud image included in the target LiDAR distortion correction information as the target anchor point information. More specifically, the target anchor point information can be obtained by interpolating the RTK positioning data before and after the first frame of the LiDAR image.

[0113] Step 105: Determine the initial heading angle and residual information based on the positioning factor map, radar coordinate information, and RTK positioning information.

[0114] In some embodiments, the executing entity can determine the initial heading angle and residual information based on the positioning factor map, the radar coordinate information, and the RTK positioning information. The specific implementation of generating the initial heading angle and its resulting technical effects can be found in step 101 of the above embodiments, and will not be repeated here. Then, the residual information can be determined using the following formula:

[0115] .

[0116] in, This indicates the residual information mentioned above.

[0117] Step 106: In response to determining that the pre-acquired observation mode information meets the preset mode conditions, coordinate transformation processing is performed on the RTK positioning information based on the target anchor point information, initial heading angle and residual information to obtain the target positioning information.

[0118] In some embodiments, the execution entity may, in response to determining that the pre-acquired observation mode information meets preset mode conditions, perform coordinate transformation processing on the RTK positioning information based on the positioning factor map, the target anchor point information, the initial heading angle, and the residual information to obtain target positioning information. Specifically, in response to determining that the pre-acquired observation mode information meets preset mode conditions, the RTK positioning information may be transformed from the W-frame to the ECEF-frame based on the first rotation matrix, the second rotation matrix, and the coordinate transformation formula included in the positioning factor map, the target anchor point information, the initial heading angle, and the residual information. The observation mode information may be determined through the following steps: when the number of observable satellites is greater than 11, it is assumed to be an open environment with good GNSS signal, where the RTK accuracy is very high; therefore, the RTK positioning result is adopted, and this mode is determined as the first mode information. When the number of observable satellites is less than 11, it can be considered that a complex urban environment has been entered. To avoid the influence of RTK positioning information with large errors on the fusion results, based on empirical values, RTK data with fewer than 9 satellites and a PDOP value greater than 3.2, fewer than 5 satellites and X and Y variances greater than 0.7 are filtered out. Then, the LIO and RTK results are fused, and this mode is determined as the second mode information. When satellite signals are interrupted or filtered out for a long time due to poor quality, the system degenerates to LIO for pose calculation, and this mode is determined as the third mode information.

[0119] The aforementioned preset mode conditions can be that the aforementioned observation mode information is the second mode information.

[0120] The above embodiments of this application have the following beneficial effects: The graph-optimized integrated positioning method of some embodiments of this application addresses the problem that RTK cannot achieve high-precision and robust positioning in complex urban environments. This paper proposes an integrated positioning algorithm. It constructs LiDAR odometry factors, IMU pre-integration factors, and RTK factors, and optimizes them using factor graphs. During the algorithm initialization phase, anchor points are obtained using interpolation, and the initial heading angle is optimized using factor graph optimization to achieve coordinate system transformation and unification. Simultaneously, an RTK filtering strategy is used to filter out poor-quality RTK information, and weights are dynamically allocated. A forced RTK addition strategy is added to prevent excessive cumulative errors caused by prolonged RTK non-fusion. Centimeter-level accuracy can be achieved in open environments, decimeter-level accuracy in complex urban environments, and stable high-precision positioning results can be provided even when RTK signals are interrupted. Future work can explore achieving a tight combination of RTK and LIO, and adding new sensors, such as visual sensors, to further improve the positioning performance of integrated navigation in complex environments.

[0121] Next reference Figures 2-5 The diagram illustrates an application scenario in an open environment (corresponding to the first mode information). For example... Figures 2-5 As shown, Figure 2 The diagram illustrates the fusion accuracy map in an open environment of some embodiments of the graph-optimized combined localization method according to this application. Figure 2 In this context, LIOR represents the graph optimization-based combined localization method of this application. RTK represents the method using only RTK localization. Time (sec) represents time (seconds). E represents East. N represents North. U represents Up. Figure 3 A table comparing root mean square (RMS) error values ​​in open environments for some embodiments of the graph-optimized combined localization method according to this application is shown. As shown in the figure, m represents meters. RMSE_E[m] represents the RMS error value corresponding to the E direction. RMSE_N[m] represents the RMS error value corresponding to the N direction. RMSE_U[m] represents the RMS error value corresponding to the U direction. RMSE_3D[m] represents the RMS error value corresponding to three-dimensional space. Figure 4 Error fluctuation diagrams in open environments are shown for some embodiments of the graph-optimized combined localization method according to this application. Figure 5 The diagram illustrates vehicle trajectory maps in an open environment, based on some embodiments of the graph-optimized combined localization method according to this application.

[0122] In open environments, RTK positioning accuracy can reach the centimeter level, making it directly usable as positioning results. Therefore, RTK is given a higher weight during the fusion process, resulting in improved accuracy such as... Figure 2As shown, the localization results of the combined algorithm are almost identical to those of RTK, meeting the accuracy requirements in open environments. The RMSE (Root Mean Square Error) of the two algorithms are as follows: Figure 3 As shown. Analysis Figure 2 It was found that the combinatorial algorithm exhibited fluctuations, which has been addressed. Figure 4 The area circled in green shows a slight fluctuation in the E direction. This is because the algorithm needs to perform time alignment before fusing RTK positioning information, i.e., interpolating the RTK coordinates of the corresponding time in the LiDAR frame. The use of interpolation introduces a small error. For example, if the vehicle is traveling at high speed in the E direction and is almost stationary in the N direction, the interpolation effect is small and almost consistent with the RTK accuracy. However, in the E direction, because the vehicle is not moving at a constant speed, the interpolation introduces some error, causing fluctuations. To verify this hypothesis, [the following text is missing from the original extract]. Figure 4 The specific route within the green box is drawn, as follows: Figure 5 As shown, green represents the starting point and red represents the ending point, and the movement is exactly in the east-west direction, so it can be proven that the fluctuation of the error is caused by the interpolation method.

[0123] Next reference Figures 6-9 The diagram illustrates an application scenario where a vehicle moves from an open area into a complex urban environment (corresponding to the second mode information). For example... Figures 6-9 As shown, Figure 6 The diagram illustrates the overall accuracy of driving from an open area into a complex urban environment, based on some embodiments of the graph-optimized combined positioning method according to this application. Figure 7 The diagram illustrates the accuracy of a vehicle traveling from an open area into a complex urban environment in accordance with some embodiments of the graph-optimized combined positioning method of this application, in the preceding open environment. Figure 8 The diagram illustrates the vehicle trajectory as it moves from an open area into a complex urban environment, according to some embodiments of the graph-optimized combined localization method of this application. Figure 9 A table comparing the root mean square error values ​​of driving from an open area into a complex urban environment is shown, based on some embodiments of the graph optimization-based combined positioning method according to this application.

[0124] Error analysis graphs obtained from the experiment, showing the transition from an open area to a complex urban environment, are as follows: Figure 6 , Figure 7 As shown, Figure 8The vehicle trajectory map shows that the accuracy of the combined positioning system in the open environment is almost identical to that of RTK. The reasons for the fluctuations were analyzed in the open environment experiment and can be verified by comparing with the trajectory map. Analyzing the latter half of the accuracy map, the observation accuracy of RTK significantly decreased, and slight signal interruptions occurred. At this point, the combined system can provide more stable and accurate positioning results. It can be seen that the LIOR algorithm shows a trend of gradually increasing error in the graph. This is because RTK data was discarded due to not meeting the filtering criteria. Without RTK data input, the system uses the LIO algorithm, leading to error accumulation. However, this situation will be resolved when RTK data is added, and the positioning accuracy can be restored based on the RTK data. The accuracy in the E, N, and U directions was improved by 46%, 50%, and 8%, respectively.

[0125] Next reference Figures 10-11 The diagram illustrates an application scenario in a complex urban environment where signal interruption occurs (i.e., corresponding to third-mode information). For example... Figures 10-11 As shown, Figure 10 The overall accuracy graph under signal interruption in a complex urban environment is shown, illustrating some embodiments of the graph-optimized combined positioning method according to this application. Figure 11 A table comparing root mean square error values ​​under signal interruption in complex urban environments is shown, illustrating some embodiments of the graph-optimized combined localization method according to this application.

[0126] The error analysis comparison chart of the combined algorithm in the transition from an open area to a complex urban environment is shown below. Figure 10 As shown, since the entire route is in a complex urban environment, the weight of RTK is reduced. Under these complex conditions, the combined algorithm maintains decimeter-level accuracy. Analysis of the latter part reveals three significant fluctuations in RTK. This is due to the reduced number of observable satellites and signal interruptions caused by vehicles entering urban canyons and obstructed parking lots. In these situations, the combined algorithm demonstrates high stability, rejecting RTK signals and using LIO for short-term positioning. Once the RTK signal quality improves, the signals are fused to correct the accumulated LIO error. Accuracy is improved by 10%, 46%, and 24% in the E, N, and U directions, respectively.

[0127] Further reference Figure 12 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a graph-optimized combined positioning device, which are similar to... Figure 1 Corresponding to the method embodiments shown, this graph-optimized combined positioning device can be specifically applied to various electronic devices.

[0128] like Figure 12As shown, a graph-optimized combined positioning device 1200 according to some embodiments includes: a generation unit 1201, a distortion correction unit 1202, a first coordinate transformation unit 1203, a first determination unit 1204, a second determination unit 1205, and a second coordinate transformation unit 1206. The generation unit 1201 is configured to generate a positioning factor map based on the historical lidar observation information, the pre-acquired historical IMU observation information, and the historical RTK positioning information, in response to determining that pre-acquired historical lidar observation information meets preset initialization conditions. The distortion correction unit 1202 is configured to perform distortion correction processing on the pre-acquired lidar observation information based on the positioning factor map to obtain target lidar distortion correction information. The first coordinate transformation unit 1203 is configured to perform coordinate transformation processing on the target lidar distortion correction information based on the positioning factor map to obtain radar coordinate information. The first determining unit 1204 is configured to determine the target anchor point information based on the aforementioned target lidar distortion correction information and the pre-acquired RTK positioning information; the second determining unit 1205 is configured to determine the initial heading angle and residual information based on the aforementioned positioning factor map, the aforementioned radar coordinate information, and the aforementioned RTK positioning information; the second coordinate transformation unit 1206 is configured to, in response to determining that the pre-acquired observation mode information satisfies the preset mode conditions, perform coordinate transformation processing on the aforementioned RTK positioning information based on the aforementioned positioning factor map, the aforementioned target anchor point information, the aforementioned initial heading angle, and the aforementioned residual information to obtain the target positioning information.

[0129] It is understandable that the units described in the graph-optimized combined positioning device 1200 and the reference Figure 1 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the graph-optimized combined positioning device 1200 and the units contained therein, and will not be repeated here.

[0130] This application also provides a computer device 1300. For example... Figure 13 As shown, computer device 1300 includes: bus 1301, processor 1302, memory 1303, and communication interface 1304. Processor 1302, memory 1303, and communication interface 1304 communicate with each other via bus 1301. Computer device 1300 can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memories in computer device 1300.

[0131] Bus 1301 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 13 The bus 1301 may be represented by a single line, but this does not mean that there is only one bus or one type of bus. The bus 1301 may include a path for transmitting information between various components of the computer device 1300 (e.g., memory 1303, processor 1302, communication interface 1304).

[0132] The processor 1302 may include any one or more processors such as a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0133] Memory 1303 may include volatile memory, such as random access memory (RAM). Memory 1303 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0134] The memory 1303 stores executable program code, and the processor 1302 executes the executable program code to implement the functions of the aforementioned generation unit, distortion removal unit, first coordinate transformation unit, first determination unit, second determination unit, and second coordinate transformation unit, thereby realizing the above-mentioned graph-optimized combined localization method. That is, the memory 1303 stores instructions for executing the above-mentioned graph-optimized combined localization method.

[0135] The communication interface 1304 uses transceiver modules such as, but not limited to, network interface cards and transceivers to enable communication between the computer device 1300 and other devices or communication networks.

[0136] This application also provides a chip, which includes a processor and a data interface. The processor reads instructions stored in the memory through the data interface to execute the above-described graph-optimized combined positioning method.

[0137] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a computing device can store, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct the computing device to perform the graph-optimized combined positioning method described above.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The above embodiments are only used 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 foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of this application.

Claims

1. A graph optimization-based combined localization method, comprising: In response to the determination that the pre-acquired historical lidar observation information meets the preset initialization conditions, a positioning factor map is generated based on the historical lidar observation information, the pre-acquired historical IMU observation information, and the historical RTK positioning information; Based on the positioning factor map, the pre-acquired lidar observation information is subjected to distortion correction processing to obtain the target lidar distortion correction information; Based on the positioning factor map, coordinate transformation processing is performed on the distortion-free information of the target lidar to obtain the lidar coordinate information; Based on the target lidar distortion correction information and the pre-acquired RTK positioning information, the target anchor point information is determined; Based on the positioning factor map, the radar coordinate information, and the RTK positioning information, the initial heading angle and residual information are determined. In response to the determination that the pre-acquired observation mode information meets the preset mode conditions, the RTK positioning information is subjected to coordinate transformation processing based on the positioning factor map, the target anchor point information, the initial heading angle and the residual information to obtain the target positioning information.

2. The graph-optimized combined localization method according to claim 1, characterized in that, The step of generating a positioning factor map based on the historical lidar observation information and pre-acquired historical IMU observation information and historical RTK positioning information includes: Based on the historical IMU observation information, the historical lidar observation information is subjected to distortion correction processing to obtain IMU pre-integration factor information and lidar observation information; Based on the lidar observation information, radar odometry factor information is constructed; Based on the historical RTK positioning information and the lidar observation information, RTK factor information is constructed; The radar odometry factor information, the IMU pre-integration factor information, and the RTK factor information are fused to obtain a positioning factor map.

3. The graph-optimized combined localization method according to claim 2, characterized in that, The historical lidar observation information includes: a lidar observation keyframe sequence, wherein the lidar observation keyframes in the lidar observation keyframe sequence include: observation timestamps; and the distortion correction processing performed on the historical lidar observation information based on the historical IMU observation information to obtain IMU pre-integration factor information and lidar observation information, including: For each lidar observation keyframe in the lidar observation keyframe sequence included in the historical lidar observation information, the following distortion correction steps are performed: The observation timestamps included in the key frames of the lidar observation are determined as the first timestamp; The observation timestamp included in the next lidar observation keyframe in the lidar observation keyframe sequence is determined as the second timestamp. Based on the first timestamp and the second timestamp, construct the IMU pre-integration factor corresponding to the historical IMU observation information; Based on the IMU pre-integration factor, generate radar prior pose information; Based on the prior pose information of the radar, distortion removal processing is performed on the key frames of the lidar observation to obtain the lidar distortion-removed frames. Each generated LiDAR distortion-corrected frame is determined as a LiDAR distortion-corrected frame sequence; Each generated IMU pre-integration factor is identified as IMU pre-integration factor information; The distortion-corrected frame sequence of the lidar is determined as lidar observation information.

4. The graph-optimized combined localization method according to claim 3, characterized in that, The historical IMU observation information includes: an IMU observation information sequence; and the step of constructing the IMU pre-integration factor corresponding to the historical IMU observation information based on the first timestamp and the second timestamp includes: The difference between the first timestamp and the second timestamp is determined as the predicted duration; Based on the predicted duration, the IMU observation information corresponding to the first timestamp in the IMU observation information sequence included in the historical IMU observation information is predicted to obtain IMU prediction information. The historical IMU observation information and the IMU prediction information are pre-integrated to obtain the IMU pre-integration factor.

5. The graph-optimized combined localization method according to claim 2, characterized in that, The lidar observation information includes: a lidar distortion-corrected frame sequence, wherein the lidar distortion-corrected frames in the lidar distortion-corrected frame sequence include: lidar point sets; and, the construction of radar odometry factor information based on the lidar observation information includes: For each LiDAR distortion-corrected frame in the LiDAR observation information sequence, the following determination steps are performed: Determine the curvature feature information corresponding to each lidar point in the lidar point set included in the lidar distortion correction frame to obtain a curvature feature information set; Based on the curvature feature information set, the lidar point set is classified to obtain the lidar planar point set and the lidar edge point set; The lidar planar point set and lidar edge point set are downsampled respectively to obtain the lidar sampling planar point set and lidar sampling edge point set; Determine the voxel map corresponding to the next LiDAR distortion frame in the LiDAR distortion frame sequence; Based on the voxel map, the planar feature distance value corresponding to each lidar sampling plane point in the lidar sampling plane point set is determined, and a planar feature distance value set is obtained; Based on the voxel map, determine the edge feature distance value corresponding to each lidar sampling edge point in the lidar sampling edge point set, and obtain the edge feature distance value set; Based on the planar feature distance value set and the edge feature distance value set, determine the pose transformation matrix; Based on the pose transformation matrix, the radar odometry factor is determined; Each generated radar odometry factor is identified as radar odometry factor information.

6. The graph-optimized combined localization method according to claim 2, characterized in that, The construction of RTK factor information based on the historical RTK positioning information and the lidar observation information includes: Based on the lidar observation information and the historical RTK positioning information, the historical anchor point information is determined; Determine the initial heading angle; Based on the historical lidar observation information and the initial heading angle, a first rotation matrix and a second rotation matrix are constructed. Based on the historical anchor point information, the first rotation matrix, and the second rotation matrix, a coordinate transformation formula is constructed. The first rotation matrix, the second rotation matrix, and the coordinate transformation formula are fused to obtain the RTK factor information.

7. A graph-optimized combined positioning device, characterized in that, include: The generation unit is configured to generate a positioning factor map based on the historical lidar observation information, the pre-acquired historical IMU observation information, and the historical RTK positioning information in response to determining that the pre-acquired historical lidar observation information meets the preset initialization conditions. The distortion correction unit is configured to perform distortion correction processing on the pre-acquired lidar observation information based on the positioning factor map to obtain the target lidar distortion correction information. The first coordinate transformation unit is configured to perform coordinate transformation processing on the distortion-free information of the target lidar based on the positioning factor map to obtain the radar coordinate information. The first determining unit is configured to determine the target anchor point information based on the target lidar distortion information and the pre-acquired RTK positioning information; The second determining unit is configured to determine the initial heading angle and residual information based on the positioning factor map, the radar coordinate information and the RTK positioning information; The second coordinate transformation unit is configured to, in response to determining that the pre-acquired observation mode information meets the preset mode conditions, perform coordinate transformation processing on the RTK positioning information based on the positioning factor map, the target anchor point information, the initial heading angle, and the residual information to obtain the target positioning information.

8. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 6.

9. A computer-readable medium, characterized in that, It stores a computer program, characterized in that the computer program, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images