A GNSS-assisted odometry-based LiDAR-IMU calibration method and system

Abstract

This invention discloses a GNSS-assisted odometry-based LiDAR-IMU calibration method and system, belonging to the field of sensor calibration technology. The method includes: acquiring the IMU's pose in real time using a GNSS / INS loose combination method; acquiring the LiDAR's pose using a LiDAR odometry calculation method; obtaining the relative pose between epochs by epoch-to-epoch difference between the two poses; constructing LiDAR-IMU rotational extrinsic parameter relationships and LiDAR-IMU translational extrinsic parameter relationships using a robot hand-eye calibration method; and based on LiDAR-IMU... The relative rotation relationship of the LiDAR-IMU is extracted from the rotation extrinsic parameter relationship, and the relative translation relationship of the LiDAR-IMU is extracted from the translation extrinsic parameter relationship. These are calculated using weighted least squares and inter-epoch iterations to obtain the exact solutions for the LiDAR-IMU rotation and translation extrinsic parameters. Based on these exact solutions, and combining the rotation and translation extrinsic parameter relationships, calibration is completed. This invention effectively suppresses the impact of IMU error drift and large errors on calibration accuracy, improving calibration accuracy and robustness in dynamic environments.

Description

Technical Field

[0001] This invention belongs to the field of sensor calibration technology, specifically relating to a GNSS-assisted odometry-based LiDAR-IMU calibration method and system. Background Technology

[0002] With the increasing demand for high-precision positioning, navigation, and perception in fields such as autonomous driving, drones, and robotics, LiDAR (Light Detection and Ranging) and Inertial Measurement Units (IMUs) have become crucial sensors for achieving high-precision spatial perception and dynamic positioning. In multi-sensor integrated systems, LiDAR-IMU calibration is a prerequisite for the effective fusion of LiDAR and IMU, and the accuracy of LiDAR-IMU calibration directly affects the navigation and perception accuracy of the entire system. However, improving the accuracy and reliability of the calibration process has always been a hot research topic.

[0003] Existing LiDAR-IMU calibration methods mainly fall into two categories: calibration methods based on external landmark features and calibration methods based on odometry. Calibration methods based on external landmark features rely on significant line, surface, or volumetric features in the environment, such as building corners, road markings, and planes. These methods typically require predefined environmental features and complex algorithms for extraction and matching to ensure calibration accuracy. However, the scenario requirements and calibration procedures of these methods are not user-friendly for non-experts, and their reliability and applicability are limited in dynamic or feature-sparse environments. Odometry-based LiDAR-IMU calibration methods, on the other hand, calculate the LiDAR pose and IMU pose separately using the platform's motion, and perform calibration by correlating their poses. This method does not rely on external environmental features and has strong adaptability. However, the IMU's bias error (gyroscope bias, accelerometer deviation) directly affects the accuracy of the calibration results, especially during long-term motion, where the accumulation of bias error can lead to deviations in the calibration results.

[0004] Therefore, it is necessary to design a LiDAR-IMU calibration method and system based on odometry that uses Global Navigation Satellite System (GNSS) to improve positioning accuracy in order to address the above problems. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the prior art by providing a GNSS-assisted odometry-based LiDAR-IMU calibration method and system. By employing a GNSS / INS loose combination method to acquire the IMU pose in real time, and using the LiDAR odometry method to acquire the LiDAR pose, the solution is obtained using weighted least squares and inter-epoch iteration. This method is not only simple to operate and effectively suppresses the impact of IMU error drift and large errors on calibration accuracy, but also significantly improves the calibration accuracy and system robustness in dynamic environments.

[0006] According to one aspect of this specification, a GNSS-assisted odometry-based LiDAR-IMU calibration method is provided, comprising:

[0007] The IMU pose is acquired in real time using a GNSS / INS loose combination method, and the LiDAR pose is acquired using the LiDAR odometry method. The relative poses between the two poses are obtained by epoch difference.

[0008] Based on the obtained relative poses between their respective epochs, the robot hand-eye calibration method is used to construct the LiDAR-IMU rotational extrinsic parameter relationship and the LiDAR-IMU translational extrinsic parameter relationship.

[0009] The relative rotation relationship of LiDAR-IMU is extracted based on the rotation extrinsic parameter relationship of LiDAR-IMU, and the relative translation relationship of LiDAR-IMU is extracted based on the translation extrinsic parameter relationship of LiDAR-IMU. The exact solutions of LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters are obtained by weighted least squares and inter-epoch iteration, respectively.

[0010] The calibration was completed based on the exact solutions of the LiDAR-IMU rotational and translational extrinsic parameters, combined with the LiDAR-IMU rotational and translational extrinsic parameter relationships.

[0011] Furthermore, the IMU pose is acquired in real time using a loose combination of GNSS / INS methods, wherein the pose information of the carrier to be calibrated is measured in real time by the built-in sensor of the IMU.

[0012] Furthermore, the LiDAR-IMU rotational extrinsic parameter relationship and the LiDAR-IMU translational extrinsic parameter relationship are constructed using a robot hand-eye calibration method, including:

[0013] The robot hand-eye calibration method is used to correlate the relative poses between epochs to obtain the robot hand-eye calibration relationship;

[0014] The pose in the robot hand-eye calibration equation is converted into a combination of rotation and translation, and the LiDAR-IMU rotational extrinsic parameter equation and the LiDAR-IMU translational extrinsic parameter equation are calculated.

[0015] Furthermore, based on the relative rotation relationship of LiDAR-IMU, the solution is obtained through weighted least squares and inter-epoch iteration, including:

[0016] The rotation matrix in the relative rotation relationship of LiDAR-IMU is converted into a quaternion, which is then solved by weighted least squares.

[0017] A system of rotational equations is constructed based on K time points. The system of rotational equations is solved through inter-epoch iterations to obtain the precise values ​​of the rotational extrinsic parameters of the LiDAR-IMU.

[0018] Furthermore, based on the relative translation relationship of LiDAR-IMU, the solution is obtained through weighted least squares and inter-epoch iteration, including:

[0019] The coordinate system origin in the relative translation relationship of LiDAR-IMU is translated, and the solution is obtained by weighted least squares.

[0020] A system of translation equations is constructed based on K time points. The system of translation equations is solved through inter-epoch iterations to obtain the precise values ​​of the LiDAR-IMU rotational extrinsic parameters.

[0021] According to one aspect of this specification, a GNSS-assisted odometry-based LiDAR-IMU calibration system is provided, comprising:

[0022] The pose acquisition module is used to acquire IMU pose in real time using the GNSS / INS loose combination method; the LiDAR pose is acquired using the LiDAR odometry method, and the two poses are respectively obtained by inter-epoch difference to obtain their respective inter-epoch relative poses.

[0023] A module for constructing extrinsic parameter relationships is used to construct LiDAR-IMU rotational extrinsic parameter relationships and LiDAR-IMU translational extrinsic parameter relationships based on the obtained relative poses between their respective epochs using robot hand-eye calibration methods.

[0024] The extrinsic parameter solving module is used to extract the relative rotation relationship of LiDAR-IMU based on the LiDAR-IMU rotation extrinsic parameter relationship, and to extract the relative translation relationship of LiDAR-IMU based on the LiDAR-IMU translation extrinsic parameter relationship. The calculation is performed by weighted least squares and inter-epoch iteration to obtain the exact solutions of LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters respectively.

[0025] The calibration module is used to complete the calibration based on the exact solutions of the LiDAR-IMU rotational extrinsic parameters and the LiDAR-IMU translational extrinsic parameters, combined with the LiDAR-IMU rotational extrinsic parameter relations and the LiDAR-IMU translational extrinsic parameter relations.

[0026] According to one aspect of this specification, an electronic device is provided, including a memory and a processor, the memory storing a computer program, characterized in that the processor executes the computer program to implement the steps of the GNSS-assisted odometry-based LiDAR-IMU calibration method.

[0027] According to one aspect of this specification, a computer-readable storage medium is provided having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the GNSS-assisted odometry-based LiDAR-IMU calibration method.

[0028] Compared with the prior art, the beneficial effects of the present invention are:

[0029] 1. The present invention proposes a GNSS-assisted odometry-based LiDAR-IMU calibration method and system. By introducing GNSS assistance, the IMU pose is calculated using a loose combination of GNSS / INS. Compared with the traditional method that only uses IMU recursion, it effectively suppresses IMU error drift. The odometry-based method for LiDAR-IMU calibration does not require complex external features and is easy to operate.

[0030] 2. The present invention proposes a GNSS-assisted LiDAR-IMU calibration method and system based on odometry. By constructing the pose relationship of the LiDAR-IMU extrinsic parameters and solving it using weighted least squares and inter-epoch iteration, the method effectively suppresses the influence of large errors on calibration accuracy and significantly improves the calibration accuracy and system robustness in dynamic environments. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0032] Figure 1 This is a flowchart of the LiDAR-IMU calibration method according to an embodiment of the present invention;

[0033] Figure 2 This is a schematic diagram of the external parameters of the LiDAR-IMU in an embodiment of the present invention;

[0034] Figure 3 This is a schematic diagram of GNSS / INS and LiDAR trajectories according to an embodiment of the present invention;

[0035] Figure 4 This is a diagram showing the LiDAR-IMU rotational extrinsic parameter estimation curves according to an embodiment of the present invention.

[0036] Figure 5 This is a LiDAR-IMU translational extrinsic parameter estimation curve from an embodiment of the present invention. Detailed Implementation

[0037] 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.

[0038] This invention provides a GNSS-assisted odometry-based LiDAR-IMU calibration method, such as... Figure 1 As shown, the process includes: acquiring IMU pose in real time using a GNSS / INS loose combination method, acquiring LiDAR pose using LiDAR odometry, and obtaining the relative pose between each epoch through inter-epoch difference; based on the obtained relative poses between epochs, constructing LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters using a robot hand-eye calibration method; extracting the relative rotation relationship of LiDAR-IMU based on the LiDAR-IMU rotation extrinsic parameters and the relative translation relationship based on the LiDAR-IMU translation extrinsic parameters, respectively, and calculating them through weighted least squares and inter-epoch iteration to obtain the exact solutions for the LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters; and completing the calibration by combining the exact solutions of the LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters with the LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters.

[0039] Specifically, suppose the pose The pose of coordinate system a in coordinate system b is: ,in This represents rotation along the three coordinate axes. This represents the translation of the origin of the coordinate system. Let the extrinsic parameters of the LiDAR-IMU be... Where I represents the IMU coordinate system and L represents the LiDAR coordinate system, calibrating the LiDAR-IMU means determining... The value of the LiDAR-IMU extrinsic parameter is obtained as follows: Figure 2 As shown.

[0040] Specifically, the LiDAR pose is calculated using the open-source LiDAR odometry method A-LOAM, and is represented in the world coordinate system (World, W frame) as follows: ; The pose is calculated using a loose combination of GNSS / INS methods, that is, combining GNSS RTK position results with IMU data, and represented in the navigation coordinate system (N frame) as follows: The W-series and N-series definitions differ, and due to the existence of LiDAR-IMU extrinsic parameters, the LiDAR trajectory and IMU trajectory do not overlap, such as... Figure 3 As shown. Although the LiDAR and IMU trajectories do not overlap, their shapes are still similar.

[0041] Specifically, set for Tie The conversion of systems, posture and posture The transformation relationship can be written as:

[0042] (1)

[0043] For two consecutive epochs The pose relationship between LiDAR and IMU can be written in the following form:

[0044] (2)

[0045] (3)

[0046] Inverting both sides of equation (2), we get:

[0047] (4)

[0048] Multiplying equation (4) by the left and right sides of equation (3) respectively, we get:

[0049]

[0050] (5)

[0051]

[0052] This eliminates the unknown quantity. , obtained the shape of Hand-eye alignment formula. Position Write as rotation Peaceful relocation By combining the forms, we can obtain:

[0053] (6)

[0054] Expanding equation (6), deriving it, and extracting the relationships corresponding to rotation and translation, we can obtain:

[0055] (7)

[0056] (8)

[0057] Equations (7) and (8) are derived from the hand-eye calibration relationship and include LiDAR-IMU rotational extrinsic parameters. Translation of external parameters The relation is also the solution. and The foundation.

[0058] Specifically, according to equation (7), let , The two represent consecutive epochs, respectively. The LiDAR and IMU are rotated relative to each other. The right side of equation (7) is then rotated. Moving it to the left yields:

[0059] (9)

[0060] Equation (9) is also The form is a hand-eye calibration relation for rotation. Rotation matrix The special orthogonal group SO(3) on the manifold is not easy to solve directly, so the rotation matrix is ​​used. Write as quaternions From the form, we can obtain:

[0061] (10)

[0062] in, , represent identity matrix Representing quaternions The real part, Representing quaternions Composed of the imaginary part vector, Representative vector Construct an antisymmetric matrix.

[0063] According to equation (10), At each moment, the following system of equations can be constructed:

[0064] (11)

[0065] in, Representing the The weights of each observation equation are calculated as follows: , Represents an exponential function. The scaling factor is denoted by quaternion. Solving using weighted least squares yields:

[0066] (12)

[0067] Specifically, in the actual solution process, each additional epoch of observation, i.e., the expansion of the equation system... And solving it again, we get:

[0068] (13)

[0069] The relative rotation relationship of LiDAR-IMU is accurately solved through inter-epoch iteration. After the solution is obtained, Convert to rotation matrix Or, the three Euler angles can be recovered.

[0070] Specifically, each solution will be... Substitute into equation (8) and rearrange to obtain two consecutive epochs. The translation relationship between them is as follows:

[0071] (14)

[0072] According to equation (14), At each moment, the following system of equations can be constructed:

[0073] (15)

[0074] in, ,

[0075] Where the weight is Translation Solving using weighted least squares yields:

[0076] (16)

[0077] Specifically, in the actual solution, the relative rotation of the LiDAR-IMU is solved in each iteration. Subsequently, new translational observation equations were added based on historical observations, resulting in the following set of equations:

[0078] (17)

[0079] Solving by interepoch iteration ,get The exact solution.

[0080] This invention also provides LiDAR-IMU extrinsic parameter estimation curves, such as... Figure 4 , Figure 5 As shown, the rotation angle converges in approximately 10 seconds, and the translation converges within 20 seconds, demonstrating that the method of this invention can rapidly estimate the rotational and translational extrinsic parameters of LiDAR-IMU. Furthermore, from... Figure 4 , Figure 5 It can also be seen that after the rotation and translation converge, a stable state can be reached over a long period of time, indicating that a stable and reliable calibration can be achieved. The embodiments of this invention also evaluate the IMU-LiDAR extrinsic parameters using the root mean square error (rms) and standard deviation (std). As shown in Table 1, the LiDAR-IMU extrinsic parameter calibration proposed in this invention has high accuracy, with a rotation angle statistical error better than 0.1° and a translation statistical error in the decimeter range.

[0081]

[0082] This invention also provides a GNSS-assisted odometry-based LiDAR-IMU calibration system, comprising the following steps: In the pose acquisition module, the IMU pose is acquired in real time using a GNSS / INS loose combination method; the LiDAR pose is acquired using a LiDAR odometry method, and the two poses are respectively obtained by inter-epoch difference to obtain their respective inter-epoch relative poses; then, in the extrinsic parameter relationship construction module, the LiDAR-IMU rotation extrinsic parameter relationship and the LiDAR-IMU translation extrinsic parameter relationship are constructed using a robot hand-eye calibration method; based on the constructed LiDAR-IMU rotation... The relative rotation and relative translation relationships of the LiDAR-IMU are extracted using extrinsic parameter relationships and LiDAR-IMU translation extrinsic parameter relationships. These relationships are then calculated using weighted least squares and inter-epoch iterations to obtain the exact solutions for the LiDAR-IMU rotation and translation extrinsic parameters. Finally, by combining the LiDAR-IMU rotation and translation extrinsic parameter relationships, the LiDAR-IMU calibration is completed. This effectively suppresses the impact of large errors on calibration accuracy and significantly improves the calibration accuracy and system robustness in dynamic environments.

[0083] Based on the same inventive concept as the foregoing embodiments, this embodiment of the invention also provides an electronic device, including a memory and a processor. The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the GNSS-assisted odometry-based LiDAR-IMU calibration method proposed in the above embodiments.

[0084] This invention also provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, this program is used to suppress IMU error drift and the impact of large errors on calibration accuracy, improving calibration accuracy and system robustness in dynamic environments. The storage medium can be any non-volatile storage device such as a hard disk, solid-state drive, flash drive, or optical disk, used to store computer program code and necessary data files. The stored computer program includes: a pose acquisition module, an extrinsic parameter relationship construction module, an extrinsic parameter solution module, and a calibration module.

[0085] Finally, it should be noted that the above specific embodiments are merely representative examples of the present invention. Obviously, the present invention is not limited to the above specific embodiments and many variations are possible. Any simple modifications, equivalent changes, and alterations made to the above specific embodiments based on the technical essence of the present invention should be considered within the protection scope of the present invention.

Claims

1. A GNSS-assisted odometry-based LiDAR-IMU calibration method, characterized in that, include: The IMU pose is acquired in real time using a GNSS / INS loose combination method, and the LiDAR pose is acquired using the LiDAR odometry method. The relative poses between the two poses are obtained by epoch difference. Based on the obtained relative poses between epochs, the LiDAR-IMU rotational extrinsic parameter relationships and LiDAR-IMU translational extrinsic parameter relationships are constructed using a robot hand-eye calibration method; among them, the LiDAR-IMU rotational extrinsic parameter relationship is as follows: The LiDAR-IMU translation extrinsic parameter relationship is as follows: L represents the LiDAR coordinate system, W represents the world coordinate system, N represents the navigation coordinate system, I represents the IMU coordinate system, R represents the rotation extrinsic parameter, and t represents the translation extrinsic parameter. The relative rotation relationship of LiDAR-IMU is extracted based on the rotation extrinsic parameter relationship of LiDAR-IMU, and the relative translation relationship of LiDAR-IMU is extracted based on the translation extrinsic parameter relationship of LiDAR-IMU. The exact solutions of LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters are obtained by weighted least squares and inter-epoch iteration, respectively. The calibration was completed based on the exact solutions of the LiDAR-IMU rotational and translational extrinsic parameters, combined with the LiDAR-IMU rotational and translational extrinsic parameter relationships.

2. The GNSS-assisted odometry-based LiDAR-IMU calibration method according to claim 1, characterized in that, The rotational and translational extrinsic parameter relationships of LiDAR-IMU were constructed using a robot hand-eye calibration method, including: The robot hand-eye calibration method is used to correlate the relative poses between epochs to obtain the robot hand-eye calibration relationship; The pose in the robot hand-eye calibration equation is converted into a combination of rotation and translation, and the LiDAR-IMU rotational extrinsic parameter equation and the LiDAR-IMU translational extrinsic parameter equation are calculated.

3. The GNSS-assisted odometry-based LiDAR-IMU calibration method according to claim 1, characterized in that, Based on the relative rotation relationship of LiDAR-IMU, the solution is obtained through weighted least squares and inter-epoch iteration, including: The rotation matrix in the relative rotation relationship of LiDAR-IMU is converted into a quaternion, which is then solved by weighted least squares. A system of rotational equations is constructed based on K time points. The system of rotational equations is solved through inter-epoch iterations to obtain the precise values ​​of the rotational extrinsic parameters of the LiDAR-IMU.

4. A GNSS-assisted odometry-based LiDAR-IMU calibration method according to claim 1, characterized in that, Based on the relative translation relationship of LiDAR-IMU, the solution is obtained through weighted least squares and inter-epoch iteration, including: The coordinate system origin in the relative translation relationship of LiDAR-IMU is translated, and the solution is obtained by weighted least squares. A system of translation equations is constructed based on K time points. The system of translation equations is solved through inter-epoch iterations to obtain the precise values ​​of the translation extrinsic parameters of LiDAR-IMU.

5. A GNSS-assisted odometry-based LiDAR-IMU calibration system, characterized in that, include: The pose acquisition module is used to acquire IMU pose in real time using the GNSS / INS loose combination method; the LiDAR pose is acquired using the LiDAR odometry method, and the two poses are respectively obtained by inter-epoch difference to obtain their respective inter-epoch relative poses. An extrinsic parameter relationship module is constructed to generate LiDAR-IMU rotational and translational extrinsic parameter relationships based on the obtained relative poses between epochs, using a robot hand-eye calibration method. The LiDAR-IMU rotational extrinsic parameter relationship is as follows: The LiDAR-IMU translation extrinsic parameter relationship is as follows: L represents the LiDAR coordinate system, W represents the world coordinate system, N represents the navigation coordinate system, I represents the IMU coordinate system, R represents the rotation extrinsic parameter, and t represents the translation extrinsic parameter. The extrinsic parameter module is used to extract the relative rotation relationship of LiDAR-IMU based on the rotation extrinsic parameter relationship and the relative translation relationship of LiDAR-IMU based on the translation extrinsic parameter relationship. The calculation is performed by weighted least squares and inter-epoch iteration to obtain the exact solutions of LiDAR-IMU rotation extrinsic parameters and LiDAR-IMU translation extrinsic parameters respectively. The calibration module is used to complete the calibration based on the exact solutions of the LiDAR-IMU rotational extrinsic parameters and the LiDAR-IMU translational extrinsic parameters, combined with the LiDAR-IMU rotational extrinsic parameter relations and the LiDAR-IMU translational extrinsic parameter relations.

6. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the GNSS-assisted odometry-based LiDAR-IMU calibration method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the GNSS-assisted odometry-based LiDAR-IMU calibration method according to any one of claims 1 to 4.

Images