Positioning method, device, electronic device and storage medium for autonomous driving vehicle

Abstract

The present application discloses a positioning method, device, electronic device, and storage medium for an autonomous driving vehicle, the method comprising: when the autonomous driving vehicle is about to enter a preset positioning area, obtaining a laser point cloud map corresponding to the preset positioning area sent from the cloud, the preset positioning area being an area without available satellite positioning signals, and the laser point cloud map being constructed based on high-precision inertial navigation data; obtaining the current IMU positioning data, thereby locally filtering the laser point cloud map to obtain a local laser point cloud map; obtaining the current laser point cloud data, and determining the current laser odometer posture in combination with the local laser point cloud map; based on the current laser odometer posture, optimizing using a preset graph optimization algorithm to obtain an optimized posture, thereby performing fusion positioning. The present application uses a laser point cloud map constructed based on high-precision inertial navigation data for positioning, with high positioning accuracy, and improves the real-time positioning by locally filtering the laser point cloud map.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a positioning method, device, electronic device, and storage medium for an autonomous driving vehicle. Background Art

[0002] Simultaneous Localization and Mapping (SLAM) technology can accurately achieve environmental mapping, positioning, and multi-point navigation. The current SLAM technology can be divided into laser SLAM (Lidar SLAM) and visual SLAM (Visual SLAM). The sensor used by laser SLAM is laser radar, while visual SLAM uses depth camera. Laser SLAM technology is relatively mature, has less error, and is sufficient for use in the current environment.

[0003] However, maps built based on existing laser SLAM technology are difficult to apply to vehicle positioning in some special scenarios such as tunnels, and their positioning accuracy cannot meet the requirements of autonomous driving vehicles. The SLAM positioning solution based on ICP (Iterative Closest Point) and NDT (Normal Distributions Transform) global matching has a low output frequency in scenarios such as tunnels, cannot meet the requirements of real-time positioning, and has poor accuracy. Summary of the invention

[0004] The embodiments of the present application provide a positioning method, device, electronic device, and storage medium for an autonomous driving vehicle to improve the positioning accuracy and real-time positioning of the autonomous driving vehicle in specific scenarios.

[0005] The present application embodiment adopts the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a positioning method for an autonomous driving vehicle, wherein the method comprises:

[0007] When the autonomous driving vehicle is about to enter a preset positioning area, a laser point cloud map corresponding to the preset positioning area sent by the cloud is obtained, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data;

[0008] Acquire current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map;

[0009] Acquire current laser point cloud data, and determine the current laser odometer posture according to the current laser point cloud data and the local laser point cloud map;

[0010] Based on the current laser odometer posture, a preset graph optimization algorithm is used to perform optimization to obtain an optimized posture, and fusion positioning is performed according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

[0011] Optionally, when the autonomous driving vehicle is about to enter a preset positioning area, obtaining a laser point cloud map corresponding to the preset positioning area sent by the cloud includes:

[0012] When the autonomous driving vehicle is about to enter a preset positioning area, determining a positioning state of a satellite positioning signal;

[0013] When the positioning state of the satellite positioning signal is in an available state, obtaining corresponding inertial navigation positioning data;

[0014] The inertial navigation positioning data is sent to the cloud, and a laser point cloud map of a preset positioning area corresponding to the inertial navigation positioning data sent by the cloud is received.

[0015] Optionally, the laser point cloud map includes a corner feature point cloud map and a surface feature point cloud map, and the locally filtering the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map includes:

[0016] Performing local filtering on the corner feature point cloud map according to the current IMU positioning data to obtain a corner feature local point cloud map;

[0017] The surface feature point cloud map is locally filtered according to the current IMU positioning data to obtain a surface feature local point cloud map.

[0018] Optionally, the current IMU positioning data is current latitude and longitude information calculated by the IMU, and the locally filtering the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map includes:

[0019] Based on the current longitude and latitude information, the laser point cloud map is intercepted to obtain a local laser point cloud map block of a preset size as the local laser point cloud map.

[0020] Optionally, the optimizing based on the current laser odometer posture by using a preset graph optimization algorithm to obtain an optimized posture includes:

[0021] Determining factors of a preset map optimization algorithm, the factors comprising at least one of an IMU pre-integration factor and a wheel speed factor;

[0022] Optimization is performed according to the current laser odometer posture and the factors of the preset graph optimization algorithm to obtain an optimized posture.

[0023] Optionally, before optimizing based on the current laser odometer posture using a preset graph optimization algorithm, the method further includes:

[0024] Acquire the initial reference point information sent by the cloud, wherein the initial reference point information includes latitude, longitude and height information and point cloud data of the initial reference point;

[0025] Determine the transformation relationship between the laser radar coordinate system and the world coordinate system according to the initial reference point information;

[0026] The optimized posture is a posture in a laser radar coordinate system, and the fusion positioning is performed according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle, including:

[0027] Based on the transformation relationship between the laser radar coordinate system and the world coordinate system, the posture in the laser radar coordinate system is transformed to the world coordinate system to obtain the posture in the world coordinate system.

[0028] Optionally, the optimizing based on the current laser odometer posture by using a preset graph optimization algorithm to obtain an optimized posture includes:

[0029] Obtaining a status identifier of a preset positioning area sent by the cloud;

[0030] The optimized posture is marked according to the state identifier of the preset positioning area, so as to perform fusion positioning according to the marked optimized posture.

[0031] In a second aspect, an embodiment of the present application further provides a positioning device for an autonomous driving vehicle, wherein the device comprises:

[0032] A first acquisition unit is used to acquire a laser point cloud map corresponding to a preset positioning area sent by the cloud when the autonomous driving vehicle is about to enter the preset positioning area, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data;

[0033] A local filtering unit, used to obtain current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map;

[0034] A first determination unit, configured to obtain current laser point cloud data, and determine a current laser odometer posture based on the current laser point cloud data and the local laser point cloud map;

[0035] A fusion positioning unit is used to optimize the current laser odometer posture using a preset graph optimization algorithm to obtain an optimized posture, and perform fusion positioning according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

[0036] In a third aspect, an embodiment of the present application further provides an electronic device, including:

[0037] Processor; and

[0038] A memory arranged to store computer executable instructions which, when executed, cause the processor to perform any of the methods described above.

[0039] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple application programs, the electronic device executes any of the aforementioned methods.

[0040] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: the positioning method of the autonomous driving vehicle in the embodiment of the present application first obtains the laser point cloud map corresponding to the preset positioning area sent by the cloud when the autonomous driving vehicle is about to enter the preset positioning area, and the preset positioning area is an area without available satellite positioning signals; then obtains the current IMU positioning data, and performs local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map; then obtains the current laser point cloud data, and determines the current laser odometer posture according to the current laser point cloud data and the local laser point cloud map; finally, based on the current laser odometer posture, optimizes using the preset graph optimization algorithm to obtain the optimized posture, and performs fusion positioning according to the optimized posture to obtain the fusion positioning result of the autonomous driving vehicle. The positioning method of the autonomous driving vehicle in the embodiment of the present application uses the laser point cloud map constructed based on high-precision inertial navigation data to locate the autonomous driving vehicle, has high positioning accuracy, and by performing local filtering on the laser point cloud map, compared with the global matching method, greatly improves the real-time positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0042] Figure 1 A schematic diagram of a flow chart of a positioning method for an autonomous driving vehicle in an embodiment of the present application;

[0043] Figure 2 This is a schematic diagram of the structure of a positioning device for an autonomous driving vehicle in an embodiment of the present application;

[0044] Figure 3 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0046] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0047] The present application embodiment provides a method for positioning an autonomous driving vehicle, such as Figure 1 As shown, a flow chart of a positioning method for an autonomous driving vehicle in an embodiment of the present application is provided, and the method at least includes the following steps S110 to S140:

[0048] Step S110, when the autonomous driving vehicle is about to enter a preset positioning area, obtain a laser point cloud map corresponding to the preset positioning area sent by the cloud, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data.

[0049] The positioning method of the autonomous driving vehicle of the embodiment of the present application can be executed by the vehicle side, which is mainly aimed at the real-time positioning of the autonomous driving vehicle in the preset positioning area. The preset positioning area here refers to an area without available satellite positioning signals, such as a tunnel or other similar positioning scenarios. When the autonomous driving vehicle is about to enter the preset positioning area, the laser point cloud map corresponding to the preset positioning area is obtained from the cloud, that is, before entering the preset positioning area, the laser point cloud map used for positioning is pre-loaded, which can improve the subsequent positioning efficiency and real-time positioning.

[0050] In addition, in order to ensure the accuracy of the preloaded map, the preload operation can be performed when the current position of the autonomous driving vehicle is less than a certain distance from the boundary of the preset positioning area. The specific distance can be flexibly set by those skilled in the art according to actual needs and is not specifically limited here. Of course, the map can also be loaded immediately when the autonomous driving vehicle has driven to the boundary position of the preset positioning area.

[0051] The laser point cloud map in the embodiment of the present application is constructed based on high-precision inertial navigation data collected by high-precision inertial navigation equipment. High-precision inertial navigation factors are constructed through high-precision inertial navigation data to participate in the optimization process of the preset map optimization algorithm, thereby avoiding the degradation phenomenon of laser SLAM mapping due to the small number of feature points in specific scenarios, and improving the mapping accuracy of laser SLAM.

[0052] Step S120, obtaining current IMU positioning data, and performing local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map.

[0053] The data volume of the laser point cloud map corresponding to the entire preset positioning area is very large. Directly using the traditional global matching method based on ICP or NDP algorithm has low matching efficiency. Therefore, the embodiment of the present application can further obtain the current IMU (Inertial Measurement Unit) positioning data, and then use the current IMU positioning data to perform local filtering on the laser point cloud map to obtain a local laser point cloud map, thereby reducing the data volume of the map and improving matching efficiency and positioning real-time performance.

[0054] Step S130, obtaining current laser point cloud data, and determining the current laser odometer posture according to the current laser point cloud data and the local laser point cloud map.

[0055] After obtaining the local laser point cloud map, the current laser point cloud data can be further acquired, and then the laser point cloud data is feature matched with the corresponding local laser point cloud map, so as to determine the current laser odometer posture.

[0056] Step S140, based on the current laser odometer posture, optimize using a preset graph optimization algorithm to obtain an optimized posture, and perform fusion positioning according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

[0057] Since the laser odometer posture obtained after feature matching will have a certain matching error, in order to improve the positioning accuracy, the embodiment of the present application can further use the preset graph optimization algorithm to optimize the posture based on the current laser odometer posture, so as to obtain the optimized posture. The optimized posture can be input as observation information into the Kalman filter or the extended Kalman filter for fusion positioning, thereby obtaining the fusion positioning result of the autonomous driving vehicle.

[0058] The positioning method of the autonomous driving vehicle in the embodiment of the present application uses a laser point cloud map constructed based on high-precision inertial navigation data to position the autonomous driving vehicle, which has high positioning accuracy and greatly improves the real-time positioning by performing local filtering on the laser point cloud map compared to the global matching method.

[0059] In some embodiments of the present application, when the autonomous driving vehicle is about to enter the preset positioning area, obtaining the laser point cloud map corresponding to the preset positioning area sent by the cloud includes: when the autonomous driving vehicle is about to enter the preset positioning area, determining the positioning status of the satellite positioning signal; when the positioning status of the satellite positioning signal is an available state, obtaining the corresponding inertial navigation positioning data; sending the inertial navigation positioning data to the cloud, and receiving the laser point cloud map of the preset positioning area corresponding to the inertial navigation positioning data sent by the cloud.

[0060] In the embodiment of the present application, when the autonomous driving vehicle is about to enter the preset positioning area, the positioning status of the satellite positioning signal can be determined first, because the satellite positioning signal can generally be obtained before entering the preset positioning area, but the quality of the satellite positioning signal needs to be determined. For example, it can be judged based on the RTK (Real-time kinematic) differential state of the acquired satellite positioning signal and the number of satellites. If the differential state is a fixed solution 42 and the number of satellites is >20, it can be considered that the quality of the current satellite positioning signal is good, that is, it is in an available state.

[0061] When the positioning status of the satellite positioning signal is available, the corresponding inertial navigation positioning data can be further obtained. The inertial navigation positioning data can be understood as the positioning data with higher accuracy obtained by fusing the IMU positioning data and the RTK differential data. The inertial navigation positioning data is sent to the cloud, so that the cloud can more accurately determine the current position of the autonomous driving vehicle and the preset positioning area it is about to enter, and then the laser point cloud map corresponding to the preset positioning area constructed in advance can be sent to the vehicle for subsequent positioning.

[0062] In some embodiments of the present application, the laser point cloud map includes a corner feature point cloud map and a surface feature point cloud map, and the local filtering of the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map includes: local filtering of the corner feature point cloud map according to the current IMU positioning data to obtain a corner feature local point cloud map; local filtering of the surface feature point cloud map according to the current IMU positioning data to obtain a surface feature local point cloud map.

[0063] The laser point cloud map constructed in advance in the embodiment of the present application is mainly divided into a corner feature point cloud map constructed based on corner features and a surface feature point cloud map constructed based on surface features. When performing local filtering, the corner feature point cloud map and the surface feature point cloud map can be subjected to local filtering processing respectively, thereby obtaining a corner feature local point cloud map and a surface feature local point cloud map. In the subsequent feature matching stage, the current laser point cloud data can be matched with the corner feature local point cloud map and the surface feature local point cloud map respectively, and then the two matching results are fused to obtain a laser odometer posture.

[0064] Of course, in order to further improve the matching efficiency, the embodiment of the present application may also perform overall filtering on the corner feature point cloud map and the surface feature point cloud map before performing local filtering processing, that is, reduce the data volume of the point cloud map as a whole.

[0065] In some embodiments of the present application, the current IMU positioning data is the current longitude and latitude information calculated by the IMU, and the local filtering of the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map includes: based on the current longitude and latitude information, intercepting the laser point cloud map to obtain a local laser point cloud map block of a preset size as the local laser point cloud map.

[0066] The current IMU positioning data of the embodiment of the present application may specifically refer to the current longitude and latitude information calculated by the IMU. Based on the positioning point corresponding to the longitude and latitude information, a local laser point cloud map block of a preset size can be intercepted from the laser point cloud map. For example, the positioning point corresponding to the longitude and latitude information can be used as the center point, and based on the distance of 50m up and down, left and right, and front and back of the positioning point, a local laser point cloud map block with a length, width, and height of 100m can be intercepted. Of course, the corner feature point cloud map and the surface feature point cloud map in the aforementioned embodiment can be intercepted separately here to obtain two local laser point cloud map blocks.

[0067] Therefore, the difference between local filtering and global filtering is that local filtering is to cut out a small piece of point cloud map from the overall laser point cloud map, and what changes is the size of the map, while global filtering is to filter the amount of data in the entire laser point cloud map, and what changes is the density of the points.

[0068] In addition, it should be noted that although the IMU positioning data itself will have the problem of increasing cumulative errors over time, since the embodiment of the present application only determines an approximate point cloud map block based on the position of the IMU positioning, the requirements for position accuracy are relatively low, and will not affect the subsequent matching and positioning accuracy.

[0069] In some embodiments of the present application, the optimization based on the current laser odometer posture using a preset graph optimization algorithm to obtain the optimized posture includes: determining the factors of the preset graph optimization algorithm, the factors including at least one of an IMU pre-integration factor and a wheel speed factor; optimizing according to the current laser odometer posture and the factors of the preset graph optimization algorithm to obtain the optimized posture.

[0070] The preset graph optimization algorithm of the embodiment of the present application can be implemented by factor graph optimization, so the factors of the preset graph optimization algorithm can be determined first, for example, IMU pre-integration factor, wheel speed factor, etc. The pre-integration is calculated by IMU positioning data to obtain the posture transformation, and then the IMU pre-integration factor can be constructed. Based on the speed of the forward direction collected in real time by the wheel speedometer, the corresponding position at each moment is calculated. Based on these position data, the position constraint can be constructed to obtain the wheel speed factor.

[0071] Combining one or more of the above factors with the current laser odometer posture for posture optimization can improve the optimization efficiency and optimization effect, thereby improving the positioning accuracy and real-time positioning.

[0072] In some embodiments of the present application, before optimizing based on the current laser odometer posture using a preset graph optimization algorithm, the method also includes: obtaining the initial reference point information sent by the cloud, the initial reference point information including the latitude, longitude and height information and point cloud data of the initial reference point; determining the transformation relationship between the lidar coordinate system and the world coordinate system based on the initial reference point information; the optimized posture is the posture in the lidar coordinate system, and performing fusion positioning based on the optimized posture to obtain the fusion positioning result of the autonomous driving vehicle includes: based on the transformation relationship between the lidar coordinate system and the world coordinate system, transforming the posture in the lidar coordinate system to the world coordinate system to obtain the posture in the world coordinate system.

[0073] Since the laser point cloud map constructed in advance in the embodiment of the present application is established in the laser radar coordinate system, the optimized posture obtained based on this is also in the laser radar coordinate system. In the autonomous driving scenario, it is often necessary to determine the absolute position and posture of the vehicle in the world coordinate system. Therefore, there is a need to convert the optimized posture into the world coordinate system.

[0074] Based on this, the embodiment of the present application can obtain the initial reference point information from the cloud. The initial reference point here can be understood as a positioning point determined by the autonomous driving vehicle before entering the preset positioning area. Therefore, its absolute position in the world coordinate system, that is, the longitude and latitude information and its corresponding point cloud data in the lidar coordinate system can be known. Then, based on the information of these two dimensions, the transformation relationship of rotation and translation between the lidar coordinate system and the world coordinate system can be calculated, and then based on the transformation relationship, the posture in the lidar coordinate system can be converted to the world coordinate system.

[0075] In some embodiments of the present application, the current laser odometer posture is optimized using a preset graph optimization algorithm to obtain the optimized posture, including: obtaining a status identifier of the preset positioning area sent by the cloud; marking the optimized posture according to the status identifier of the preset positioning area, so as to perform fusion positioning according to the marked optimized posture.

[0076] Since the preset positioning areas in the embodiments of the present application are mainly aimed at some special positioning scenarios, the cloud defines corresponding status identifiers for different preset positioning areas during the map construction process. For example, if it is a tunnel scenario, there is a unique identifier for the tunnel positioning area, which is used to distinguish different positioning areas and sources of positioning results.

[0077] Based on this, in the optimization and fusion positioning stage of the posture, the embodiment of the present application can obtain the state identifier defined by the cloud for the preset positioning area, and then annotate the positioning result based on the state identifier to characterize the positioning scenario in which the positioning result was obtained. Since the positioning result obtained by the embodiment of the present application is of high accuracy, its corresponding confidence is higher than that of the positioning result without the state identifier in the same or similar positioning scenario. Therefore, in the fusion positioning stage, it can be given a higher confidence according to the state identifier, thereby improving the accuracy and effect of the fusion positioning.

[0078] In order to facilitate the understanding of the various embodiments of the present application, the construction process of the laser point cloud map of the present application is further introduced, which specifically includes: obtaining mapping data collected by the vehicle in a preset positioning area, the preset positioning area is an area without available satellite positioning signals, and the mapping data includes high-precision inertial navigation data and laser point cloud data; post-processing the high-precision inertial navigation data to obtain post-processed high-precision inertial navigation data, and the post-processed high-precision inertial navigation data includes high-precision inertial navigation posture; constructing factors of a preset map optimization algorithm based on the post-processed high-precision inertial navigation data and the laser point cloud data, the factors of the preset map optimization algorithm include laser odometer factors and high-precision inertial navigation factors; optimizing based on the factors of the preset map optimization algorithm to obtain an optimized posture, and constructing a map based on the optimized posture.

[0079] The laser SLAM mapping method of the embodiment of the present application can be executed by the cloud. When performing laser SLAM mapping, it is necessary to first obtain the mapping data collected by the vehicle in the preset positioning area. The preset positioning area here refers to the area where the satellite positioning signal is not available, and the number of feature points in these areas is relatively small. The above-mentioned mapping data mainly includes high-precision inertial navigation data and laser point cloud data collected in the entire preset positioning area. In order to avoid the degradation phenomenon caused by the small number of feature points extracted from the laser point cloud data in specific scenarios, the embodiment of the present application can use the high-precision inertial navigation equipment installed on the vehicle to collect high-precision inertial navigation positioning data. Compared with traditional inertial navigation equipment, it can reduce the cumulative error of positioning data that increases over time.

[0080] After that, the high-precision inertial navigation data needs to be post-processed to further improve its accuracy. The post-processing operation here mainly solves the high-precision inertial navigation data collected by the high-precision inertial navigation equipment to obtain data such as position, attitude and speed.

[0081] The embodiment of the present application can use a preset graph optimization algorithm to optimize the pose of the map, and specifically can use a factor graph optimization method. Therefore, after obtaining the post-processed high-precision inertial navigation data, it is necessary to construct the corresponding high-precision inertial navigation factor based on the post-processed high-precision inertial navigation data, and the corresponding laser radar odometer factor can be constructed based on the laser point cloud data as the basis for subsequent optimization. In addition, since the frequency of output data from different sensors is different, in order to facilitate subsequent processing, a certain time synchronization algorithm such as interpolation method can also be used here to synchronize the post-processed high-precision inertial navigation data and laser point cloud data.

[0082] Based on the above laser odometer factor and high-precision inertial navigation factor, the pose can be optimized to obtain the optimized pose, and then the laser point cloud data can be spliced ​​based on the optimized pose to obtain a constructed laser point cloud map. By introducing the optimization constraint of the high-precision inertial navigation factor, the degradation of the laser odometer factor due to the small number of feature points can be avoided, reducing the mapping error.

[0083] The above-mentioned laser SLAM mapping process optimizes the existing laser SLAM mapping scheme for a specific positioning area. By constructing a high-precision inertial navigation factor to participate in the optimization process of the preset map optimization algorithm, it avoids the degradation phenomenon of laser SLAM mapping due to the small number of feature points in specific scenarios, thereby improving the mapping accuracy of laser SLAM.

[0084] In some embodiments of the present application, the post-processed high-precision inertial navigation data also includes IMU data, and constructing the factors of the preset graph optimization algorithm based on the post-processed high-precision inertial navigation data and the laser point cloud data includes: performing IMU pre-integration processing based on the IMU data to obtain an IMU pre-integration factor as a factor of the preset graph optimization algorithm; optimizing the factors based on the preset graph optimization algorithm to obtain the optimized posture includes: optimizing based on the IMU pre-integration factor, the laser odometer factor and the high-precision inertial navigation factor to obtain the optimized posture.

[0085] The post-processed high-precision inertial navigation data of the embodiment of the present application also includes IMU data, where the IMU data may refer to the IMU data after the errors such as zero bias and table zero bias are removed. The pre-integration can be calculated through the IMU data to obtain the posture transformation, so that the IMU pre-integration factor can be constructed. The IMU pre-integration factor, the laser odometer factor, and the high-precision inertial navigation factor are used as factors of the preset graph optimization algorithm for back-end optimization to obtain the optimized posture, thereby further improving the optimization efficiency and optimization effect.

[0086] In some embodiments of the present application, the optimization based on the IMU pre-integration factor, the laser odometer factor and the high-precision inertial navigation factor to obtain the optimized posture includes: adjusting the posture confidence of the IMU pre-integration factor, the laser odometer factor and the high-precision inertial navigation factor based on the high-precision inertial navigation factor; optimizing according to the posture confidence of the adjusted IMU pre-integration factor, the posture confidence of the adjusted laser odometer factor and the posture confidence of the adjusted high-precision inertial navigation factor to obtain the optimized posture.

[0087] When the embodiment of the present application performs posture optimization based on the IMU pre-integration factor, laser odometer factor and high-precision inertial navigation factor, the covariances corresponding to different factors are different, and thus the corresponding posture confidences in the optimization process are also different. The larger the covariance, the lower the posture confidence, and the smaller the covariance, the higher the corresponding posture confidence.

[0088] Since the high-precision inertial navigation factor of the embodiment of the present application is constructed based on the high-precision inertial navigation data collected by the high-precision inertial navigation device, it does not depend on the number and richness of the feature points extracted from the environment, and there is no problem of cumulative error increasing over time. Therefore, its corresponding covariance should be smaller, and thus it can be given a higher pose confidence during the pose optimization process, while reducing the pose confidence corresponding to the IMU pre-integration factor and the laser odometer factor. That is, compared with the IMU pre-integration factor and the laser odometer factor, there is a greater probability of believing the pose corresponding to the high-precision inertial navigation factor.

[0089] The above process is essentially that in the process of posture optimization, the high-precision inertial navigation factor constructed based on the above embodiment adaptively adjusts the covariance size or confidence size of different optimization factors, so that factors with higher precision have higher weights in the optimization process, and factors with lower precision have relatively smaller weights in the optimization process, which is more suitable for mapping scenarios with a small number of feature points such as tunnels.

[0090] In addition, it should be noted that since the covariance size of the IMU pre-integration factor is mainly affected by time accumulation, the adjustment of the confidence size of the IMU pre-integration factor can be gradually reduced as time accumulates.

[0091] In some embodiments of the present application, the post-processed high-precision inertial navigation data also includes IMU data, and the factors of constructing a preset graph optimization algorithm based on the post-processed high-precision inertial navigation data and the laser point cloud data include: using the IMU data to dedistort the laser point cloud data to obtain dedistorted laser point cloud data; extracting corner point features and surface features from the dedistorted laser point cloud data; and constructing the laser odometer factor based on the extracted corner point features and surface features.

[0092] When constructing the laser odometry factor, the embodiment of the present application can first perform de-distortion correction processing on the laser point cloud data. Specifically, the IMU data corresponding to the laser point cloud data of the current frame can be used to calculate the rotation increment, and the IMU pre-integration can be used to calculate the translation increment. Then, the laser point at each moment of the frame of laser point cloud data can be corrected for motion distortion. At the same time, the attitude angle of the IMU data and the posture corresponding to the IMU pre-integration can be used to roughly initialize the posture of the laser point cloud data of the current frame, thereby completing the de-distortion processing of the laser point cloud data.

[0093] After that, the curvature of each point of the laser point cloud data after motion distortion correction is calculated, and then the corner point features and surface features can be extracted based on the size of the curvature. The position and posture of the laser point cloud data are calculated based on the corner point features and surface features to construct the laser odometer factor.

[0094] In some embodiments of the present application, before optimizing the factors based on the preset graph optimization algorithm to obtain the optimized posture, the method also includes: performing loop detection based on the laser point cloud data; constructing a loop factor based on the loop detection result as a factor of the preset graph optimization algorithm.

[0095] When implementing laser SLAM mapping in specific scenarios, the embodiments of the present application can further improve the optimization efficiency and optimization effect by adding loop constraints. In the early stage, the vehicle can be controlled to drive along the loop trajectory, and the mapping data collected can be used for subsequent loop detection. Specifically, the existing loop detection algorithm such as the bag-of-words model or the deep learning-based method can be used for loop detection, so that the posture constraint relationship can be established based on the loop detection results, and the posture optimization can be performed in combination with other factors to improve the posture optimization efficiency and optimization effect.

[0096] In some embodiments of the present application, the mapping data also includes wheel speed data. Before optimizing based on the factors of the preset map optimization algorithm to obtain the optimized posture, the method also includes: determining the position data corresponding to each frame of wheel speed data based on the wheel speed data; constructing a wheel speed factor according to the position data corresponding to each frame of wheel speed data as a factor of the preset map optimization algorithm.

[0097] In specific mapping scenarios such as tunnels, the wheel speed data collected by the wheel speed meter on the vehicle is less affected by external factors and has higher accuracy. Therefore, the embodiment of the present application can further construct a wheel speed factor based on the wheel speed data, thereby providing richer constraint information for subsequent posture optimization.

[0098] Specifically, in the mapping scenario, it can be assumed that the direction-finding speed and radial speed in the wheel speed data collected by the tachometer are both 0, that is, there is only the speed in the forward direction. Since the initial position of the vehicle before entering the preset positioning area is known, the speed in the forward direction collected in real time by the wheel speedometer can be used to calculate the position corresponding to each moment. Based on these position data, position constraints can be constructed, thereby providing more reference information for the subsequent optimization process, further improving the optimization efficiency and effect.

[0099] In some embodiments of the present application, after optimizing the factors based on the preset map optimization algorithm to obtain the optimized pose, and constructing the map according to the optimized pose, the method also includes: comparing the optimized pose with the high-precision inertial navigation pose; determining the mapping accuracy of the optimized pose according to the comparison result; and returning the mapping accuracy of the optimized pose to the vehicle end, so that the vehicle end determines whether re-mapping is needed according to the mapping accuracy of the optimized pose.

[0100] In order to evaluate the quality of mapping, after obtaining the optimized posture, the embodiment of the present application can also compare the optimized posture used for mapping with the high-precision inertial navigation data. Since the high-precision inertial navigation data has a high accuracy, its corresponding posture data can be used as a basis for judging the error size of the optimized posture.

[0101] The mapping accuracy of the optimized posture can be determined based on the comparison results. For example, if the horizontal angle error in the attitude angle is less than 0.1, the heading angle error is less than 0.2, and the corresponding statistical error is less than twice the standard deviation (2sigma), the position error is less than 0.15m, and the corresponding statistical error is less than twice the standard deviation (2sigma), then the accuracy is considered to be high. In this way, the mapping results can be scored, or other forms can be used to indicate whether the mapping results meet the preset accuracy requirements, and the results can be returned to the vehicle side, which can use this to determine whether secondary mapping is needed.

[0102] The present application also provides a positioning device 200 for an autonomous driving vehicle, such as Figure 2 As shown, a schematic diagram of the structure of a positioning device for an autonomous driving vehicle in an embodiment of the present application is provided, wherein the device 200 at least includes: a first acquisition unit 210, a local filtering unit 220, a first determination unit 230, and a fusion positioning unit 240, wherein:

[0103] The first acquisition unit 210 is used to acquire a laser point cloud map corresponding to a preset positioning area sent by the cloud when the autonomous driving vehicle is about to enter the preset positioning area, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data;

[0104] A local filtering unit 220 is used to obtain current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map;

[0105] A first determination unit 230 is used to obtain current laser point cloud data, and determine a current laser odometer posture according to the current laser point cloud data and the local laser point cloud map;

[0106] The fusion positioning unit 240 is used to optimize the current laser odometer posture using a preset graph optimization algorithm to obtain an optimized posture, and perform fusion positioning according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

[0107] In some embodiments of the present application, the first acquisition unit 210 is specifically used to: determine the positioning status of the satellite positioning signal when the autonomous driving vehicle is about to enter a preset positioning area; obtain corresponding inertial navigation positioning data when the positioning status of the satellite positioning signal is available; send the inertial navigation positioning data to the cloud, and receive a laser point cloud map of the preset positioning area corresponding to the inertial navigation positioning data sent by the cloud.

[0108] In some embodiments of the present application, the laser point cloud map includes a corner feature point cloud map and a surface feature point cloud map, and the local filtering unit 220 is specifically used to: perform local filtering on the corner feature point cloud map according to the current IMU positioning data to obtain a corner feature local point cloud map; perform local filtering on the surface feature point cloud map according to the current IMU positioning data to obtain a surface feature local point cloud map.

[0109] In some embodiments of the present application, the current IMU positioning data is the current longitude and latitude information calculated by the IMU, and the local filtering unit 220 is specifically used to: based on the current longitude and latitude information, intercept the laser point cloud map to obtain a local laser point cloud map block of a preset size as the local laser point cloud map.

[0110] In some embodiments of the present application, the fusion positioning unit 240 is specifically used to: determine the factors of a preset map optimization algorithm, the factors including at least one of an IMU pre-integration factor and a wheel speed factor; optimize according to the current laser odometer posture and the factors of the preset map optimization algorithm to obtain an optimized posture.

[0111] In some embodiments of the present application, the device also includes: a second acquisition unit, used to acquire the initial reference point information sent by the cloud, the initial reference point information including the latitude, longitude and height information and point cloud data of the initial reference point; a second determination unit, used to determine the transformation relationship between the lidar coordinate system and the world coordinate system based on the initial reference point information; the optimized posture is the posture in the lidar coordinate system, and the fusion positioning unit is specifically used to: based on the transformation relationship between the lidar coordinate system and the world coordinate system, transform the posture in the lidar coordinate system to the world coordinate system to obtain the posture in the world coordinate system.

[0112] In some embodiments of the present application, the fusion positioning unit 240 is specifically used to: obtain the status identifier of the preset positioning area sent by the cloud; mark the optimized posture according to the status identifier of the preset positioning area, and perform fusion positioning according to the marked optimized posture.

[0113] It can be understood that the positioning device of the above-mentioned autonomous driving vehicle can implement the various steps of the positioning method of the autonomous driving vehicle provided in the aforementioned embodiments. The relevant explanations about the positioning method of the autonomous driving vehicle are applicable to the positioning device of the autonomous driving vehicle and will not be repeated here.

[0114] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 3 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.

[0115] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0116] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0117] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a positioning device for the autonomous driving vehicle at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations:

[0118] When the autonomous driving vehicle is about to enter a preset positioning area, a laser point cloud map corresponding to the preset positioning area sent by the cloud is obtained, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data;

[0119] Acquire current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map;

[0120] Acquire current laser point cloud data, and determine the current laser odometer posture according to the current laser point cloud data and the local laser point cloud map;

[0121] Based on the current laser odometer posture, a preset graph optimization algorithm is used to perform optimization to obtain an optimized posture, and fusion positioning is performed according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

[0122] The above application Figure 1 The method performed by the positioning device of the autonomous driving vehicle disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or instructions in the form of software. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0123] The electronic device may also perform Figure 1 A method for executing a positioning device of an autonomous driving vehicle in a vehicle positioning system Figure 1 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0124] The present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by an electronic device including multiple application programs, enable the electronic device to execute Figure 1 The method performed by the positioning device of the autonomous driving vehicle in the illustrated embodiment is specifically used to perform:

[0125] When the autonomous driving vehicle is about to enter a preset positioning area, a laser point cloud map corresponding to the preset positioning area sent by the cloud is obtained, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data;

[0126] Acquire current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map;

[0127] Acquire current laser point cloud data, and determine the current laser odometer posture according to the current laser point cloud data and the local laser point cloud map;

[0128] Based on the current laser odometer posture, a preset graph optimization algorithm is used to perform optimization to obtain an optimized posture, and fusion positioning is performed according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

[0129] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0130] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0131] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0133] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0134] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0135] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0136] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0137] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0138] The above is only the embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A positioning method for an autonomous driving vehicle, wherein: The method comprises: When the autonomous driving vehicle is about to enter a preset positioning area, a laser point cloud map corresponding to the preset positioning area sent by the cloud is obtained, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data; Acquire current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map; Acquire current laser point cloud data, and determine the current laser odometer posture according to the current laser point cloud data and the local laser point cloud map; Based on the current laser odometer posture, a preset graph optimization algorithm is used to perform optimization to obtain an optimized posture, and fusion positioning is performed according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

2. The method of claim 1, wherein: When the autonomous driving vehicle is about to enter the preset positioning area, obtaining the laser point cloud map corresponding to the preset positioning area sent by the cloud includes: When the autonomous driving vehicle is about to enter a preset positioning area, determining a positioning state of a satellite positioning signal; When the positioning state of the satellite positioning signal is in an available state, obtaining corresponding inertial navigation positioning data; The inertial navigation positioning data is sent to the cloud, and a laser point cloud map of a preset positioning area corresponding to the inertial navigation positioning data sent by the cloud is received.

3. The method of claim 1, wherein: The laser point cloud map includes a corner feature point cloud map and a surface feature point cloud map, and the local filtering of the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map includes: Performing local filtering on the corner feature point cloud map according to the current IMU positioning data to obtain a corner feature local point cloud map; The surface feature point cloud map is locally filtered according to the current IMU positioning data to obtain a surface feature local point cloud map.

4. The method of claim 1, wherein: The current IMU positioning data is the current latitude and longitude information calculated by the IMU, and the local filtering of the laser point cloud map according to the current IMU positioning data to obtain the local laser point cloud map includes: Based on the current longitude and latitude information, the laser point cloud map is intercepted to obtain a local laser point cloud map block of a preset size as the local laser point cloud map.

5. The method of claim 1, wherein: The optimized posture obtained by optimizing the current laser odometer posture using a preset graph optimization algorithm includes: Determining factors of a preset map optimization algorithm, the factors comprising at least one of an IMU pre-integration factor and a wheel speed factor; Optimization is performed according to the current laser odometer posture and the factors of the preset graph optimization algorithm to obtain an optimized posture.

6. The method of claim 1, wherein: Before optimizing based on the current laser odometer posture using a preset graph optimization algorithm, the method further includes: Acquire the initial reference point information sent by the cloud, wherein the initial reference point information includes latitude, longitude and height information and point cloud data of the initial reference point; Determine the transformation relationship between the laser radar coordinate system and the world coordinate system according to the initial reference point information; The optimized posture is a posture in a laser radar coordinate system, and the fusion positioning is performed according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle, including: Based on the transformation relationship between the laser radar coordinate system and the world coordinate system, the posture in the laser radar coordinate system is transformed to the world coordinate system to obtain the posture in the world coordinate system.

7. The method of claim 1, wherein: The optimized posture obtained by optimizing the current laser odometer posture using a preset graph optimization algorithm includes: Obtaining a status identifier of a preset positioning area sent by the cloud; The optimized posture is marked according to the state identifier of the preset positioning area, so as to perform fusion positioning according to the marked optimized posture.

8. A positioning device for an autonomous driving vehicle, wherein: The device comprises: A first acquisition unit is used to acquire a laser point cloud map corresponding to a preset positioning area sent by the cloud when the autonomous driving vehicle is about to enter the preset positioning area, wherein the preset positioning area is an area without available satellite positioning signals, and the laser point cloud map is constructed based on high-precision inertial navigation data; A local filtering unit, used to obtain current IMU positioning data, and perform local filtering on the laser point cloud map according to the current IMU positioning data to obtain a local laser point cloud map; A first determination unit, configured to obtain current laser point cloud data, and determine a current laser odometer posture based on the current laser point cloud data and the local laser point cloud map; A fusion positioning unit is used to optimize the current laser odometer posture using a preset graph optimization algorithm to obtain an optimized posture, and perform fusion positioning according to the optimized posture to obtain a fusion positioning result of the autonomous driving vehicle.

9. An electronic device, comprising: processor; as well as A memory arranged to store computer executable instructions, which when executed cause the processor to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, causes the electronic device to execute any one of the methods of claims 1 to 7.

Images