Positioning method and device applied to automatic driving, intelligent vehicle and storage medium

Abstract

This invention provides a positioning method and device, an intelligent vehicle, and a storage medium for autonomous driving. The positioning method for autonomous driving includes: Step S100: Real-time acquisition of the current frame's LiDAR point cloud data Ft+1; Step S200: Determining whether the LiDAR point cloud data Ft+1 is keyframe data; Step S300: If the LiDAR point cloud data Ft+1 is determined to be keyframe data, sampling and extracting feature points from the LiDAR point cloud data Ft+1, constructing an objective function by iterating the nearest point and adjacent keyframes, and obtaining the LiDAR odometry pose estimation value; if the LiDAR point cloud data Ft+1 is determined not to be keyframe data, discarding the LiDAR point cloud data Ft+1; Step S400: Fusing the LiDAR odometry pose estimation value and the LiDAR odometry pose estimation value to update the LiDAR odometry pose. This embodiment of the invention, while maintaining real-time performance, more effectively utilizes LiDAR point cloud data, improving positioning accuracy.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle technology, and more specifically, to a tightly coupled positioning method for LiDAR wheel speedometers applied to autonomous driving, a fresh air control device for intelligent vehicle cooling mode, an intelligent vehicle, and a readable storage medium. Background Technology

[0002] In autonomous driving technology, accurate vehicle positioning is the foundation for ensuring the stability and reliability of the entire system, and a basic prerequisite for realizing advanced autonomous driving functions.

[0003] In the field of autonomous driving, vision-based or LiDAR-based localization solutions are already quite mature. Although vision-based localization solutions have certain advantages in system cost and loop closure detection, the sensitivity of vision sensors to lighting conditions, rapid movement, and initialization may pose certain risks to the stability of the system.

[0004] To better achieve vehicle localization in different environments, LiDAR-based localization systems have become the mainstream solution for addressing vehicle localization issues in the field of autonomous driving. Currently available high-resolution LiDAR systems can capture environmental information over long distances and large areas, unaffected by conditions such as nighttime.

[0005] Because autonomous driving systems need to deal with complex environmental conditions, LiDAR-based positioning methods often need to be fused with other sensors to further enhance the robustness of the positioning system.

[0006] LiDAR odometry typically achieves localization by calculating the relative pose transformation between two consecutive frames using scanning matching methods (such as ICP and GICP). However, due to computational efficiency issues, full-point-cloud matching methods have been largely replaced by feature-based matching methods in many real-time-critical applications. Therefore, optimizing LiDAR localization algorithms hinges on balancing real-time performance with more efficient use of LiDAR point cloud data to improve positioning accuracy. Summary of the Invention

[0007] Therefore, this invention provides a tightly coupled positioning method for LiDAR wheel speedometers applied to autonomous driving, which improves positioning accuracy by making more effective use of LiDAR point cloud data while taking into account real-time performance.

[0008] To address the aforementioned problems, this invention provides a tightly coupled positioning method for LiDAR wheel speedometers applied to autonomous driving, comprising: Step S100: acquiring the current frame LiDAR point cloud data Ft+1 in real time; Step S200: determining whether the LiDAR point cloud data Ft+1 is keyframe data; Step S300: if the LiDAR point cloud data Ft+1 is determined to be keyframe data, sampling and extracting feature points from the LiDAR point cloud data Ft+1, constructing an objective function by iterating the nearest point and adjacent keyframes, and obtaining the LiDAR odometry pose estimation value; if the LiDAR point cloud data Ft+1 is determined not to be keyframe data, discarding the LiDAR point cloud data Ft+1; Step S400: fusing the LiDAR odometry pose estimation value and the LiDAR odometry pose estimation value to update the LiDAR odometry pose.

[0009] Compared with existing technologies, the technical effects achieved by this solution are as follows: On the one hand, this embodiment, through a tightly coupled positioning framework based on LiDAR and wheel speedometer, can more effectively utilize environmental information around the vehicle compared to a positioning scheme using a single LiDAR sensor, and greatly reduce the impact of the surrounding environment on the positioning results, thereby improving the robustness of the algorithm and solving the vehicle positioning problem for autonomous driving in different environments; on the other hand, this embodiment, based on a keyframe strategy and LiDAR odometry using feature points, greatly reduces the computational load of point clouds while maintaining positioning accuracy.

[0010] In one embodiment of the present invention, step S200 specifically includes: step S210: calculating the current pose of the radar odometry of the current frame lidar point cloud data Ft+1. Step S220: Calculate the previous time-state pose of the radar odometry for the previous frame of lidar point cloud data Ft. Step S230: Using the current pose and the pose at the previous time step The pose increment [Δw, Δh] is obtained by subtraction, where Δw is the component of the rotation increment in the yaw direction and Δh is the absolute value of the translation increment; Step S240: When Δw is greater than the rotation threshold and Δh is greater than the translation threshold, the current frame lidar point cloud data Ft+1 is key frame data; if Δw is less than or equal to the rotation threshold and / or Δh is less than or equal to the translation threshold, the current frame lidar point cloud data Ft+1 is not key frame data.

[0011] Compared with existing technologies, the technical effects achieved by this solution are as follows: the corresponding pose increment is obtained through rotation increment and translation increment, and the rotation threshold, translation threshold and rotation increment are compared. Only when Δw is greater than the rotation threshold and Δh is greater than the translation threshold is the current frame LiDAR point cloud data Ft+1 used as key frame data. All others are not key frame data, thereby improving the accuracy of key frame identification.

[0012] In one embodiment of the present invention, the calculation of the pose of the radar odometer of the lidar point cloud data in steps S210 and S220 specifically includes: reading wheel speed pulse data, integrating the wheel speed data according to the kinematic equation, and estimating the pose increment of the lidar odometer on the planar component.

[0013] Compared with existing technologies, the technical effects achieved by this solution are as follows: by reading wheel speed pulse data and integrating the wheel speed meter data according to the kinematic equation, the pose increment of the lidar odometer on the planar component is estimated, thereby improving the accuracy of the pose of the lidar odometer in calculating lidar point cloud data.

[0014] In one embodiment of the present invention, the step of reading wheel speed pulse data, integrating the wheel speed meter data according to the kinematic equation, and estimating the pose increment of the lidar odometer in the planar component specifically includes: at the current time t, the state of the wheel speed odometer is o. t =[w t ,v t ], w t The instantaneous angular velocity of the current wheel speed gauge, and v t The instantaneous linear velocity of the current wheel speedometer is given, and Where, k l ,k r v represents the wheel speed coefficients for the left and right wheels. l ,v r Let L be the linear velocity of the left and right wheels, and L be the wheelbase of the left and right wheels; at time t, the wheel speedometer pose is... The wheel velocity sensor pose at time t+1 is estimated based on the kinematic equations. in, in, This represents the rotation matrix from the wheel velocity gauge coordinate system to the world frame; This represents the translation matrix from the wheel velocity gauge coordinate system to the world system; Δt is the time interval from time t to time t+1. Let be a Euclidean group, and let represent the extrinsic parameter matrices from the wheel velocity gauge coordinate system to the world system at times t and t+1, respectively.

[0015] Compared with existing technologies, the technical effects achieved by this solution are as follows: (Based on the wheel speed odometer status...) t and wheel speed gauge position and posture By obtaining the pose increment, the accuracy of the positioning algorithm can be improved and the impact of the cumulative error of the lidar can be reduced.

[0016] In one embodiment of the present invention, the step of sampling and extracting feature points from the lidar point cloud data Ft+1 includes: SOR filtering for noise reduction: setting a threshold based on the average distance of the k neighboring points of each point in the lidar point cloud data Ft+1 to remove discrete noise points.

[0017] Compared with existing technologies, the technical effect achieved by this solution is: by using SOR filtering to remove discrete noise points, the accuracy of the positioning algorithm is improved.

[0018] In one embodiment of the present invention, the step of sampling and extracting feature points from the lidar point cloud data Ft+1 further includes: ISS key point extraction: setting a search radius r, establishing a local coordinate system in the spatial neighborhood of each point r in the denoised point cloud data, calculating the corresponding covariance matrix based on the distances from all points in the spatial neighborhood to the key point, obtaining its eigenvalues, setting a threshold and removing points that do not meet the threshold, and denoting the sampled key frame as F′. t+1 The frame preceding the keyframe is denoted as F′. t .

[0019] Compared with existing technologies, the technical effects achieved by this solution are as follows: by extracting key points through ISS, obtaining their feature values, setting thresholds and removing points that do not meet the thresholds, the computational load of point clouds is greatly reduced while maintaining positioning accuracy.

[0020] In one embodiment of the present invention, the construction of the objective function by iterating over the nearest point and adjacent keyframes includes: [f(F′)] t and F′ t+1 The pose is solved and estimated using the iterative nearest point method, and the objective function is as follows: Where, p i , p′ i F′ t+1 F′ t The i-th pair of feature points in the data, F′ t To F′ t+1 The initial value for the iterative transformation relationship.

[0021] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: through the analysis of F′ t and F′ t+1 F′ is obtained by solving and estimating the pose using the iterative nearest point method.t To F′ t+1 The initial values ​​of the transformation relationship are used to facilitate the updating of the pose of the lidar odometry.

[0022] On the other hand, the present invention also provides a positioning device for autonomous driving, comprising: an acquisition module for acquiring current frame LiDAR point cloud data Ft+1; a judgment module for judging whether the LiDAR point cloud data Ft+1 is keyframe data; and a control module, wherein if the control module judges that the LiDAR point cloud data Ft+1 is keyframe data, it samples and extracts feature points from the LiDAR point cloud data Ft+1, constructs an objective function by iterating the nearest point and adjacent keyframes, and obtains the LiDAR odometry pose estimation value; if the control module judges that the LiDAR point cloud data Ft+1 is not keyframe data, it discards the LiDAR point cloud data Ft+1; and fuses the LiDAR odometry pose estimation value with the LiDAR odometry pose estimation value to update the LiDAR odometry pose.

[0023] The positioning device for autonomous driving in this embodiment is used to implement the tightly coupled positioning method of lidar wheel speedometer for autonomous driving as described in any embodiment of the present invention. Therefore, it has all the beneficial effects of the tightly coupled positioning method of lidar wheel speedometer for autonomous driving as described in any embodiment of the present invention, which will not be repeated here.

[0024] In another aspect, embodiments of the present invention also provide an intelligent vehicle, including: a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein when the program or instructions are executed by the processor, they implement the steps of the tightly coupled positioning method for lidar wheel speedometers applied to autonomous driving as described in any of the above embodiments.

[0025] The intelligent vehicle in this embodiment operates as the tightly coupled positioning method of lidar wheel speedometer applied to autonomous driving according to any embodiment of the present invention. Therefore, it has all the beneficial effects of the tightly coupled positioning method of lidar wheel speedometer applied to autonomous driving according to any embodiment of the present invention, which will not be repeated here.

[0026] In another aspect, embodiments of the present invention also provide a readable storage medium, the readable storage medium including a stored computer program, wherein, when the computer program is run by a processor, it controls the device where the storage medium is located to perform the steps of the lidar wheel speed meter tightly coupled positioning method for autonomous driving as described in any embodiment of the present invention.

[0027] The readable storage medium in this embodiment includes a stored computer program, and the computer program, when run by a processor, controls the device where the storage medium is located to perform the steps of the tightly coupled positioning method of lidar wheel speedometer for autonomous driving as described in any embodiment of the present invention. Therefore, it has all the beneficial effects of the tightly coupled positioning method of lidar wheel speedometer for autonomous driving as described in any embodiment of the present invention, which will not be repeated here.

[0028] By adopting the technical solution of the present invention, the following technical effects can be achieved:

[0029] (1) This embodiment uses a tightly coupled positioning framework based on lidar and wheel speed meter. Compared with the positioning scheme of a single lidar sensor, it can make more effective use of the environmental information around the vehicle and greatly reduce the impact of the surrounding environment on the positioning results, improve the robustness of the algorithm, and solve the vehicle positioning problem of autonomous driving in different environments.

[0030] (2) This embodiment is based on key frame strategy and feature point lidar odometry, which greatly reduces the amount of point cloud computation while taking into account positioning accuracy.

[0031] (3) Status of wheel speed and odometer t and wheel speed gauge position and posture By obtaining the pose increment, the accuracy of the positioning algorithm can be improved and the impact of the cumulative error of the lidar can be reduced. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings to be used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart of the tightly coupled positioning method of lidar wheel speedometer applied to autonomous driving as described in the first embodiment of the present invention;

[0034] Figure 2 This is a schematic block diagram of the positioning device applied to autonomous driving according to the second embodiment of the present invention;

[0035] Figure 3 This is a block diagram of the intelligent vehicle according to the third embodiment of the present invention;

[0036] Figure 4 This is a schematic diagram of the structure of the readable storage medium according to the fourth embodiment of the present invention.

[0037] Explanation of reference numerals in the attached figures:

[0038] 100 is a positioning device used in autonomous driving; 110 is an acquisition module; 120 is a judgment module; 130 is a control module; 200 is an intelligent vehicle; 210 is a memory; 211 is a computer program; 230 is a processor; 300 is a readable storage medium; 310 is a computer-executable instruction. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] [First Embodiment]

[0041] See Figure 1 The tightly coupled positioning method of LiDAR wheel speedometer applied to autonomous driving specifically includes:

[0042] Step S100: Real-time acquisition of the current frame's LiDAR point cloud data Ft+1;

[0043] Step S200: Determine whether the lidar point cloud data Ft+1 is keyframe data;

[0044] Step S300: If the lidar point cloud data Ft+1 is determined to be key frame data, the lidar point cloud data Ft+1 is sampled and feature points are extracted. The objective function is constructed by iterating the nearest point and adjacent key frames, and the lidar odometry pose estimation value is obtained. If the lidar point cloud data Ft+1 is determined not to be key frame data, the lidar point cloud data Ft+1 is discarded.

[0045] Step S400: Fuse the LiDAR odometry pose estimate with the LiDAR odometry pose estimate to update the LiDAR odometry pose.

[0046] For example, on the one hand, this embodiment, through a tightly coupled positioning framework based on LiDAR and wheel speedometer, can more effectively utilize the environmental information around the vehicle compared to a positioning scheme based on a single LiDAR sensor, and greatly reduce the impact of the surrounding environment on the positioning results, thereby improving the robustness of the algorithm and solving the vehicle positioning problem in autonomous driving under different environments; on the other hand, this embodiment, based on a keyframe strategy and LiDAR odometry based on feature points, greatly reduces the computational load of point clouds while maintaining positioning accuracy.

[0047] Specifically, step S200 includes:

[0048] Step S210: Calculate the current pose of the radar odometry for the current frame of LiDAR point cloud data Ft+1.

[0049] Step S220: Calculate the previous time-state pose of the radar odometry for the previous frame of lidar point cloud data Ft.

[0050] Step S230: Using the current pose pose at the previous moment Subtracting them yields the pose increment [Δw, Δh], where Δw is the component of the rotation increment in the yaw direction, and Δh is the absolute value of the translation increment.

[0051] Step S240: When Δw is greater than the rotation threshold and Δh is greater than the translation threshold, the current frame LiDAR point cloud data Ft+1 is key frame data. If Δw is less than or equal to the rotation threshold and / or Δh is less than or equal to the translation threshold, the current frame LiDAR point cloud data Ft+1 is not key frame data.

[0052] For example, the corresponding pose increment is obtained by rotation increment and translation increment. The rotation threshold, translation threshold and rotation increment are compared. Only when Δw is greater than the rotation threshold and Δh is greater than the translation threshold is the current frame LiDAR point cloud data Ft+1 used as key frame data. All others are not key frame data, thereby improving the accuracy of key frame identification.

[0053] In one specific embodiment, the rotation threshold is 10° and the translation threshold is 1m.

[0054] Furthermore, the calculation of the pose of the radar odometer in steps S210 and S220 of the lidar point cloud data specifically includes: reading wheel speed pulse data, integrating the wheel speed data according to the kinematic equation, and estimating the pose increment of the lidar odometer on the planar component.

[0055] For example, by reading wheel speed pulse data and integrating the wheel speed meter data according to the kinematic equation, the pose increment of the lidar odometer in the planar component can be estimated, thereby improving the accuracy of the lidar odometer pose in calculating lidar point cloud data.

[0056] Preferably, wheel speed pulse data is read, and the wheel speed meter data is integrated according to the kinematic equation to estimate the pose increment of the lidar odometer in the planar component. Specifically, this includes: at the current time t, the state of the wheel speed odometer is o. t =[w t ,v t ], w t The instantaneous angular velocity of the current wheel speed gauge, and vt The instantaneous linear velocity of the current wheel speedometer is given, and Where, k l ,k r v represents the wheel speed coefficients for the left and right wheels. l ,v r Let L be the linear velocity of the left and right wheels, and L be the wheelbase of the left and right wheels; at time t, the wheel speedometer pose is... The wheel velocity sensor pose at time t+1 is estimated based on the kinematic equations. in, in, This represents the rotation matrix from the wheel velocity gauge coordinate system to the world frame; This represents the translation matrix from the wheel velocity gauge coordinate system to the world system; Δt is the time interval from time t to time t+1. Let be a Euclidean group, and let represent the extrinsic parameter matrices from the wheel velocity gauge coordinate system to the world system at times t and t+1, respectively.

[0057] For example, the state of the wheel speed odometer. t and wheel speed gauge position and posture By obtaining the pose increment, the accuracy of the positioning algorithm can be improved and the impact of the cumulative error of the lidar can be reduced.

[0058] Preferably, the lidar point cloud data Ft+1 is sampled and feature points are extracted, including: SOR filtering for noise reduction: a threshold is set based on the average distance of the k neighboring points of each point in the lidar point cloud data Ft+1, and discrete noise points are removed. For example, by using SOR filtering for noise reduction, discrete noise points are removed, thereby improving the accuracy of the positioning algorithm.

[0059] Preferably, sampling and extracting feature points from the LiDAR point cloud data Ft+1 also includes: ISS key point extraction: setting a search radius r, establishing a local coordinate system in the spatial neighborhood of each point r in the denoised point cloud data, calculating the corresponding covariance matrix based on the distances from all points in the spatial neighborhood to the key point, obtaining its eigenvalues, setting a threshold and removing points that do not meet the threshold, and denoting the sampled key frame as F′. t+1 The frame preceding the keyframe is denoted as F′. t For example, by extracting key points through ISS, obtaining their feature values, setting thresholds, and removing points that do not meet the thresholds, the computational load of point clouds can be greatly reduced while maintaining positioning accuracy.

[0060] Specifically, the objective function is constructed by iterating over the nearest point and adjacent keyframes, including: [e.g., F′] t and F′ t+1 The pose is solved and estimated using the iterative nearest point method, and the objective function is as follows: Where, p i , p′ i F′ t+1 F′ t The i-th pair of feature points in the data, F′ t To F′ t+1 The initial value for the iterative transformation relationship.

[0061] For example, by analyzing F′ t and F′ t+1 F′ is obtained by solving and estimating the pose using the iterative nearest point method. t To F′ t+1 The initial values ​​of the transformation relationship are used to facilitate the updating of the pose of the lidar odometry.

[0062] Preferably, this embodiment combines wheel speed inertial odometry with lidar pose fusion optimization, specifically including:

[0063] Let the extrinsic parameter from the wheel speed odometer to the lidar sensor be T. LO Based on the wheel speed meter data, the pose at time t can be determined. Obtain the estimated pose of the lidar at time t+1. And according to The estimated pose is used to construct and solve the objective function to obtain the optimized pose. The objective function is as follows: Where, ρ r (R1,R2) represents the absolute value of the phase difference between R1 and R2 in the yaw component, ρ P (p1,p2) represents the distance between p1 and p2 in the x and y plane components.

[0064] The pose estimation given by wheel speed meter data will lack constraints in the z, roll and pitch directions. After fusing the meter data, the main improvement is made to the planar components of the lidar odometry, namely the components in the x, y and yaw directions.

[0065] [Second Embodiment]

[0066] See Figure 2This invention also provides a positioning device 100 for autonomous driving, including: an acquisition module 110 for acquiring current frame LiDAR point cloud data Ft+1; a judgment module 120 for judging whether LiDAR point cloud data Ft+1 is keyframe data; and a control module 130, which, when judging that LiDAR point cloud data Ft+1 is keyframe data, samples and extracts feature points from LiDAR point cloud data Ft+1, constructs an objective function by iterating the nearest point and adjacent keyframes, and obtains the LiDAR odometry pose estimation value; if judging that LiDAR point cloud data Ft+1 is not keyframe data, it discards LiDAR point cloud data Ft+1; and fuses the LiDAR odometry pose estimation value with the LiDAR odometry pose estimation value to update the LiDAR odometry pose.

[0067] In one specific embodiment, the acquisition module 110, judgment module 120 and control module 130 of the positioning device 100 for autonomous driving cooperate to implement the tightly coupled positioning method of lidar wheel speedometer for autonomous driving as described in the first embodiment above, which will not be repeated here.

[0068] [Third Embodiment]

[0069] See Figure 3 This is a schematic diagram of the structure of an intelligent vehicle provided in the fourth embodiment of the present invention. The intelligent vehicle 200 includes, for example, a processor 230 and a memory 210 electrically connected to the processor 230. The memory 210 stores a computer program 211. The processor 230 loads the computer program 211 to implement the lidar wheel speed meter tight coupling positioning method for autonomous driving as described in the first embodiment.

[0070] [Fourth Embodiment]

[0071] See Figure 4 This embodiment also provides a readable storage medium 300, which stores computer-executable instructions 310. When the computer-executable instructions 310 are read and executed by the processor, they control the intelligent vehicle where the readable storage medium 300 is located to implement the lidar wheel speed meter tight coupling positioning method for autonomous driving as described in the first embodiment.

[0072] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0073] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention 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; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A tightly coupled positioning method of a laser radar wheel speed meter applied to automatic driving, characterized in that, The tightly coupled positioning method of lidar wheel speedometer applied to autonomous driving includes: Step S100: Real-time acquisition of the current frame's LiDAR point cloud data Ft+1; Step S200: Determine whether the lidar point cloud data Ft+1 is keyframe data; Step S300: If the lidar point cloud data Ft+1 is determined to be key frame data, the lidar point cloud data Ft+1 is sampled and feature points are extracted. An objective function is constructed by iterating the nearest point and adjacent key frames, and the lidar odometry pose estimation value is obtained. If the lidar point cloud data Ft+1 is determined not to be key frame data, the lidar point cloud data Ft+1 is discarded. Step S400: Fuse the estimated pose values ​​of the lidar odometry and the estimated pose values ​​of the lidar odometry to update the lidar odometry pose; Step S400 specifically includes: setting the external parameters from the wheel speed odometer to the lidar sensor as follows: Based on the data from the wheel speed gauge, it is possible to... Time and position To obtain lidar in Estimated pose at time step : = And according to The estimated pose is used to construct and solve the objective function to obtain the optimized pose. The objective function is as follows: ,in, express , exist The absolute value of the difference in components. express , exist , Distance on the planar component.

2. The method of claim 1, wherein the method is applied to a laser radar wheel speed meter close-coupled positioning method for autonomous driving. Step S200 specifically includes: Step S210: calculating a current pose of a radar odometer of the current frame lidar point cloud data Ft+1 ; Step S220: calculating the last time pose of the radar odometer of the frame lidar point cloud data Ft of the last time ; Step S230: Using the current pose and the pose at the previous time step Subtraction yields the pose increment ,in, For rotation increments in Components in direction, This represents the absolute value of the translation increment. Step S240: When Greater than the rotation threshold, and When the value is greater than the translation threshold, the current frame LiDAR point cloud data Ft+1 is the key frame data. Less than or equal to the rotation threshold and / or When the translation threshold is less than or equal to the current frame lidar point cloud data Ft+1, it is not a key frame data.

3. The tightly coupled positioning method of a LIDAR wheel speed sensor for autonomous driving according to claim 2, wherein, The calculation of the radar odometry pose based on the lidar point cloud data in steps S210 and S220 specifically includes: Read wheel speed pulse data, integrate the wheel speed meter data according to the kinematic equation, and estimate the pose increment of the lidar odometer in the planar component.

4. The tightly coupled positioning method of a LIDAR wheel speed sensor for autonomous driving according to claim 3, wherein, The process of reading wheel speed pulse data, integrating the wheel speed meter data according to the kinematic equations, and estimating the pose increment of the lidar odometer in the planar component specifically includes: current At that moment, the wheel speed odometer was in the following state: , The instantaneous angular velocity of the current wheel speed gauge, and , The instantaneous linear velocity of the current wheel speedometer is given, and ; wherein, is a wheel speed coefficient for the left and right wheels, is a linear speed for the left and right wheels, is a wheelbase for the left and right wheels; Current at time t, the wheel speedometer pose is , the wheel speedometer pose at time t is estimated from the kinematic equations where , ;​ in, This represents the rotation matrix from the wheel velocity gauge coordinate system to the world frame; This represents the translation matrix from the wheel velocity gauge coordinate system to the world system; for Time's up The time interval between moments; , For Euclidean groups, respectively represent , The extrinsic parameter matrix from the time-velocity gauge coordinate system to the world system.

5. The tightly coupled positioning method of a LIDAR wheel speed sensor for autonomous driving according to claim 1, wherein, The step of sampling and extracting feature points from the lidar point cloud data Ft+1 includes: SOR filtering for noise reduction: A threshold is set based on the average distance of the k neighboring points of each point in the lidar point cloud data Ft+1 to remove discrete noise points.

6. The tightly coupled positioning method of a LIDAR wheel speed sensor for autonomous driving according to claim 1, wherein, The step of sampling and extracting feature points from the lidar point cloud data Ft+1 further includes: ISS Keypoint Extraction: In the denoised point cloud data, a local coordinate system is established within the spatial neighborhood of each point r, with a search radius r set. The covariance matrix is ​​calculated based on the distances from all points in the spatial neighborhood to the keypoint, yielding its eigenvalues. A threshold is set, and points that do not meet the threshold are removed. The sampled keyframes are denoted as... The frame preceding the keyframe is denoted as... .

7. The tightly coupled positioning method of a LIDAR wheel speed sensor for autonomous driving according to claim 6, wherein, The construction of the objective function by iterating over the nearest point and adjacent keyframes includes: To and The objective function for solving and estimating the pose by the Iterative Closest Point method is as follows: ； wherein , denotes , the first characteristic point, denotes the conversion relationship of the iteration initial value.

8. A positioning device applied to automatic driving, characterized in that, include: The acquisition module is used to acquire the current frame of lidar point cloud data Ft+1; The judgment module is used to determine whether the lidar point cloud data Ft+1 is keyframe data; The control module, when determining that the lidar point cloud data Ft+1 is keyframe data, samples and extracts feature points from the lidar point cloud data Ft+1, constructs an objective function by iterating the nearest point and adjacent keyframes, and obtains the lidar odometry pose estimation value; if it determines that the lidar point cloud data Ft+1 is not keyframe data, it discards the lidar point cloud data Ft+1; and fuses the lidar odometry pose estimation value with the lidar odometry pose estimation value to update the lidar odometry pose. The acquisition module, the judgment module, and the control module work together to implement the tightly coupled positioning method of lidar wheel speedometer for autonomous driving as described in any one of claims 1-7.

9. An intelligent vehicle, characterized by A processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the lidar wheel speed meter tightly coupled positioning method for autonomous driving as described in any one of claims 1-7.

10. A readable storage medium, characterized by, The readable storage medium includes a stored computer program, wherein, when the computer program is run by a processor, it controls the device containing the storage medium to perform the steps of the lidar wheel speedometer tightly coupled positioning method for autonomous driving as described in any one of claims 1-7.

Images