Multi-sensor calibration method and device, electronic equipment and storage medium

Abstract

The invention discloses a multi-sensor calibration method and device, electronic equipment and a storage medium. The method comprises the following steps: converting a first coordinate system according to a first conversion relation to obtain a second coordinate system; constructing candidate calibration parameters according to a transformation relation between the second coordinate system and the coordinate system of the at least one target sensor; matching a target calibration parameter from the candidate calibration parameters according to the target sensor type; and calibrating the corresponding target sensor according to the target calibration parameter. According to the method, the problems of errors, poor complex scene adaptability, insufficient precision and the like in a multi-sensor calibration process in an automatic driving vehicle are effectively solved by performing joint calibration on the multiple sensors, and the requirement of an automatic driving system on high-precision multi-sensor fusion is met.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-sensor joint calibration, and in particular to a multi-sensor calibration method, device, electronic equipment and storage medium. Background Art

[0002] In autonomous driving systems, the fusion of multi-sensor data plays a key role in achieving high-precision environmental perception, positioning and navigation. Sensors such as lidar, camera and inertial measurement unit (IMU) each provide different types of environmental information: lidar provides accurate three-dimensional point cloud data, which can be used to build high-resolution environmental maps; cameras provide rich image information, including color, texture and other details, which is suitable for object recognition and scene segmentation; IMU can provide real-time attitude and acceleration information, which is of great significance for position tracking and attitude estimation. However, these sensors vary in data type, sampling frequency, resolution and viewing angle. Therefore, in order to achieve effective fusion of multi-source information, each sensor must be calibrated with high precision.

[0003] The traditional single sensor calibration method is difficult to meet the needs of multi-sensor systems. The coordinate transformation relationship between different sensors, such as between LiDAR and camera, between LiDAR and IMU, requires precise external parameter calibration to ensure that the data after sensor fusion can be accurately aligned. In addition, the timing synchronization problem between sensors also has a great impact on the accuracy of the system. Since the sampling frequencies of LiDAR, camera and IMU are different, if the timing is inconsistent, it will cause position errors when the data is fused, thus affecting the overall stability and safety of autonomous driving. Summary of the invention

[0004] The present invention provides a multi-sensor calibration method, device, electronic equipment and storage medium to solve the problem of position deviation caused by different timings between different sensors.

[0005] According to one aspect of the present invention, a multi-sensor calibration method is provided, comprising:

[0006] The first coordinate system is transformed according to a first transformation relationship to obtain a second coordinate system; the first transformation relationship is used to characterize the transformation relationship between the first coordinate system and the reference coordinate system; the first coordinate system is a coordinate system of a reference point cloud sensor; the reference coordinate system is a coordinate system established with a preset center of the vehicle as a coordinate origin;

[0007] Constructing candidate calibration parameters according to a transformation relationship between the second coordinate system and a coordinate system of at least one target sensor; the candidate calibration parameters are used to characterize the transformation relationship between the second coordinate system and the coordinate systems of each sensor;

[0008] matching target calibration parameters from candidate calibration parameters according to the target sensor type;

[0009] The corresponding target sensor is calibrated according to the target calibration parameters.

[0010] According to another aspect of the present invention, a multi-sensor calibration device is provided, comprising:

[0011] A second coordinate system determination module, used for transforming the first coordinate system according to a first transformation relationship to obtain a second coordinate system; the first transformation relationship is used to characterize the transformation relationship between the first coordinate system and a reference coordinate system; the first coordinate system is a coordinate system of a reference point cloud sensor; the reference coordinate system is a coordinate system established with a preset center of the vehicle as a coordinate origin;

[0012] a candidate calibration parameter determination module, configured to construct candidate calibration parameters according to a transformation relationship between the second coordinate system and a coordinate system of at least one target sensor; the candidate calibration parameters are used to characterize a transformation relationship between the second coordinate system and the coordinate systems of each sensor;

[0013] a target calibration parameter determination module, for matching target calibration parameters from candidate calibration parameters according to the target sensor type;

[0014] The calibration module is used to calibrate the corresponding target sensor according to the target calibration parameters.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the multi-sensor calibration method described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the multi-sensor calibration method described in any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention transforms the first coordinate system according to the first transformation relationship to obtain the second coordinate system, which provides a transformation basis for the construction of candidate calibration parameters; constructs candidate calibration parameters according to the transformation relationship between the second coordinate system and the coordinate system of at least one target sensor, and the construction of candidate calibration parameters can provide different calibration parameters for different sensors, laying the foundation for sensor calibration; matches target calibration parameters from candidate calibration parameters according to the target sensor type; calibrates the corresponding target sensor according to the target calibration parameters, and can transform different sensors from their respective coordinate systems to the same coordinate system, solving the external parameters and timing synchronization problems between different sensors, and improving the accuracy of data fusion. This method effectively solves the problems of errors, poor adaptability to complex scenes, and insufficient accuracy in the calibration process of multiple sensors in autonomous driving vehicles by jointly calibrating multiple sensors, and meets the requirements of autonomous driving systems for high-precision multi-sensor fusion.

[0021] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A flowchart of a multi-sensor calibration method provided by an embodiment of the present invention;

[0024] Figure 2 A point cloud image in a first coordinate system provided by an embodiment of the present invention;

[0025] Figure 3 A point cloud image in a second coordinate system provided by an embodiment of the present invention;

[0026] Figure 4 A schematic diagram of a multi-sensor calibration interface provided by an embodiment of the present invention;

[0027] Figure 5 A schematic diagram of a display interface of a camera after calibration provided by an embodiment of the present invention;

[0028] Figure 6 A point cloud image in front of a first point cloud sensor provided by an embodiment of the present invention;

[0029] Figure 7 A point cloud image after a first point cloud sensor is provided in an embodiment of the present invention;

[0030] Figure 8 An image after calibration of a photographing device provided by an embodiment of the present invention;

[0031] Fig. 9 A schematic diagram of the structure of a multi-sensor calibration device provided in an embodiment of the present invention;

[0032] Fig.10 A schematic diagram of the structure of an electronic device for implementing the multi-sensor calibration method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] Figure 1 This is a flowchart of a multi-sensor calibration method provided by an embodiment of the present invention. This embodiment is applicable to the case of joint calibration of multiple sensors. The method can be executed by a multi-sensor calibration device. The multi-sensor calibration device can be implemented in the form of hardware and / or software. The multi-sensor calibration device can be configured in any electronic device with network communication function. Figure 1 As shown, the method includes:

[0036] S110. Convert the first coordinate system according to the first conversion relationship to obtain a second coordinate system.

[0037] The first conversion relationship is used to characterize the conversion relationship between the first coordinate system and the reference coordinate system.

[0038] Furthermore, the first conversion relationship can be obtained by an automatic calibration algorithm. That is, an initial conversion relationship is generated, and the first coordinate system is converted by the initial conversion relationship to obtain a reference second coordinate system. The error between the reference second coordinate system and the reference coordinate system is calculated, and the initial conversion relationship is corrected according to the error until the error is within a preset error range, thereby obtaining the first conversion relationship.

[0039] Furthermore, the first transformation relationship can also be optimized by collecting vehicle body and lidar data under different driving conditions and using an adaptive filtering method (such as extended Kalman filtering or particle filtering) to optimize the alignment parameters of the vehicle body and lidar coordinate system.

[0040] Exemplarily, the first conversion relationship can be expressed as: [x, y, z, roll, pitch, yaw].

[0041] Among them, the ground is in the plane of z=0, and xy is within the detection range; x: represents the X-axis distance between the reference point cloud sensor coordinate system and the reference coordinate system; y: represents the Y-axis distance between the reference point cloud sensor coordinate system and the reference coordinate system; z: represents the Z-axis distance between the reference point cloud sensor coordinate system and the reference coordinate system; roll: rotation around the X-axis, representing the roll angle of the reference point cloud sensor coordinate system from the reference coordinate system; pitch: rotation around the Y-axis, representing the pitch angle of the reference point cloud sensor coordinate system from the reference coordinate system; yaw: rotation around the Z-axis, representing the heading angle of the reference point cloud sensor coordinate system from the reference coordinate system.

[0042] The first coordinate system is the coordinate system of the reference point cloud sensor.

[0043] Exemplarily, the reference point cloud sensor may be a laser radar.

[0044] The reference coordinate system is a coordinate system established with the preset center of the vehicle as the coordinate origin.

[0045] The preset center may be the center of the rear axle, the center of the front axle, or the center of the vehicle body, etc.

[0046] Exemplarily, assuming that the reference coordinate system takes the center of the rear axle of the vehicle as the origin, the reference coordinate system is established by constructing the coordinate system with the origin forward x positive, left y positive, and upward z positive.

[0047] Specifically, the coordinate system of the reference point cloud sensor is transformed according to the first transformation relationship, and the coordinate system of the reference point cloud sensor is transformed into the reference coordinate system to obtain the second coordinate system.

[0048] For example, Figure 2 As shown, it is a schematic diagram of the first coordinate system; Figure 3 As shown in FIG. 1 , it is a schematic diagram of the second coordinate system. It can be seen from the figure that compared with the point cloud data of the vehicle and the objects around the vehicle obtained in the second coordinate system, the point cloud data of the vehicle and the objects around the vehicle obtained in the first coordinate system has a certain angle offset, so that the objects in the entire scene are lower in the image.

[0049] The above steps calibrate the reference point cloud sensor because in actual use, the reference point cloud sensor is not necessarily placed horizontally, and the position of the reference point cloud sensor is not the center of the vehicle. If no calibration is performed, the obtained point cloud position will deviate from the actual position of the vehicle. Calibration is to move the deviated point cloud data to the correct position and obtain the correct angle.

[0050] The above steps solve the problem of inaccurate point cloud positioning caused by the position and angle errors of the lidar, thereby ensuring the high consistency between the point cloud and the reference coordinate system.

[0051] S120: Construct candidate calibration parameters according to a transformation relationship between the second coordinate system and the coordinate system of at least one target sensor.

[0052] The candidate calibration parameters are used to characterize the transformation relationship between the second coordinate system and each sensor coordinate system.

[0053] Furthermore, the candidate calibration parameters include: a first calibration parameter, a second calibration parameter, a third calibration parameter, a fourth calibration parameter and a fifth calibration parameter.

[0054] Among them, the first calibration parameter is used to calibrate the first point cloud sensor; the second calibration parameter is used to calibrate the shooting device; the third calibration parameter is used to calibrate the second point cloud sensor; the fourth calibration parameter is used to calibrate the motion sensor; and the fifth calibration parameter is used to calibrate the positioning sensor.

[0055] Among them, the first point cloud sensor and the reference point cloud sensor are of the same type, both are laser radars. Furthermore, the reference point cloud sensor is the main laser radar and the first point cloud sensor is the secondary laser radar. The shooting device can use a camera to collect image data of the vehicle and its surrounding environment. The second point cloud sensor can use a millimeter wave radar to estimate the radial distance, arrival angle, relative radial velocity and other information of the vehicle. The motion sensor can use an Inertial Measurement Unit (Inertial Measurement Unit) to sense the acceleration, angular velocity and magnetic field strength of the vehicle, and provide the vehicle with attitude (such as pitch angle, roll angle, yaw angle), speed and position estimation. The positioning sensor can use a global satellite navigation system (such as GPS, Beidou, etc.), and the global satellite navigation system is based on RTK (Real-Time Kinematic, real-time dynamic carrier phase difference technology) to obtain vehicle positioning information with centimeter or even millimeter level accuracy in real time.

[0056] Specifically, the first calibration parameter is determined according to the conversion relationship between the reference point cloud sensor and the first point cloud sensor; the second calibration parameter is determined according to the conversion relationship between the point cloud data coordinate information and the image pixel coordinate information; the third calibration parameter is determined according to the change relationship between the point cloud data and the static object and the dynamic object; the fourth calibration parameter is determined according to the transformation relationship between the posture of the point cloud data and the motion data in different coordinate systems; the fifth calibration parameter is determined according to the association relationship between the trajectory of the point cloud data and the positioning sequence data in different coordinate systems. The obtained first calibration parameter, second calibration parameter, third calibration parameter, fourth calibration parameter and fifth calibration parameter are used to construct candidate calibration parameters.

[0057] S130 . Match target calibration parameters from candidate calibration parameters according to the target sensor type.

[0058] Specifically, target calibration parameters corresponding to the target sensor are matched from candidate calibration parameters according to the target sensor type.

[0059] S140: Calibrate the corresponding target sensor according to the target calibration parameters.

[0060] Specifically, the coordinate system corresponding to the corresponding target sensor is converted according to the target calibration parameters to complete the calibration.

[0061] For example, Figure 4 The figure shows a schematic diagram of the multi-sensor calibration interface. It can be seen from the figure that the interface contains 6 calibration buttons, namely the lidar calibration button, lidar-camera calibration button, camera intrinsic calibration button, lidar-INS calibration button, lidar-millimeter wave calibration button and lidar-positioning calibration button.

[0062] Further, when a button to be calibrated is selected through the selection device, corresponding calibration parameters are matched from candidate calibration parameters as target calibration parameters, and the sensor corresponding to the button is calibrated through the target calibration parameters.

[0063] For example, if the laser radar calibration button is selected, it indicates that the first point cloud sensor, that is, the secondary laser radar, needs to be calibrated, and the first calibration parameters are matched for calibration. Figure 5 As shown in FIG. 1 , it is a point cloud data image before calibration. From the figure, it can be seen that there is a certain angle deviation between the continuous point cloud data in gray and the continuous point cloud data in green. Figure 6 As shown in the figure, it is a calibrated point cloud data image. It can be seen from the figure that the continuous point cloud data in gray overlaps with the continuous point cloud data in green. Figure 5 compared to, Figure 6 The spiral point cloud extending outward from the coordinate center is blocked on the right side, and the point cloud shape presented by the object on the right side is also different from Figure 5 different.

[0064] For example, if the laser radar-camera calibration button is selected, the camera needs to be calibrated, and the second calibration parameter is matched for calibration. Figure 7 As shown, it is an image of the vehicle acquired by the calibrated shooting device. It can be seen from the figure that the position and distribution of the objects captured in the image are consistent with the objects described in the point cloud data, and there is no angular offset.

[0065] Optionally, the candidate calibration parameter determination process includes steps A1-A5:

[0066] Step A1: determine the first point cloud data and the second point cloud data.

[0067] The first point cloud data is data collected by the reference point cloud sensor in the second coordinate system. The first point cloud sensor is a point cloud collection sensor of the same type as the reference point cloud sensor and located in the same common view area.

[0068] Exemplarily, the reference point cloud sensor and the first point cloud sensor may be laser radars.

[0069] The second point cloud data is the point cloud data of the vehicle and the surrounding environment collected by the first point cloud sensor.

[0070] Specifically, point cloud data of the vehicle and its surroundings in the second coordinate system acquired by the reference point cloud sensor is obtained as the first point cloud data. Point cloud data of the vehicle and its surroundings acquired by the first point cloud sensor is obtained as the second point cloud data.

[0071] Step A2: voxelize the first point cloud data to obtain a reference voxel unit.

[0072] The voxelization process is used to divide the first point cloud data into grids.

[0073] Specifically, the first point cloud data is divided into grid space to obtain reference voxel units.

[0074] Exemplarily, if the first point cloud data is two-dimensional data, the first point cloud data is divided into square grids; if the first point cloud data is three-dimensional data, the first point cloud data is divided into cubic grids.

[0075] Step A3: Determine a reference density function according to the reference voxel unit and generate an initial transformation matrix.

[0076] The reference density function is used to characterize the distribution of the point cloud within the reference voxel unit.

[0077] Among them, the initial change matrix is ​​obtained by rotation and translation.

[0078] Specifically, the reference density function of each point cloud data point in the reference voxel unit is calculated, and each point in the reference voxel unit is rotated and translated to obtain an initial transformation matrix.

[0079] For example, if the data is two-dimensional, the reference density function obeys μ = the mean value σ of each point in the reference voxel unit 2 = The normal distribution of the covariance of each point in the reference voxel unit. If it is three-dimensional data, the reference density function obeys μ = the mean of the three-dimensional coordinates of each point in the reference voxel unit σ 2 =3×3 covariance matrix of the normal distribution.

[0080] Step A4: Map the second point cloud data to the reference voxel unit according to the initial transformation matrix, and determine the target density function.

[0081] The target density function is used to characterize the distribution of each point cloud in the second point cloud data within the reference voxel unit.

[0082] Specifically, the second point cloud data is transformed according to the initial transformation matrix, mapped into reference voxel units, and the target density function of the transformed second point cloud data in each reference voxel unit is calculated.

[0083] Step A5: determine a first calibration parameter according to the reference density function and the target density function.

[0084] Specifically, the initial transformation matrix is ​​optimized according to the distribution of two point cloud data represented by the reference density function and the target density function until the preset requirements are met, and the initial transformation matrix that meets the preset requirements is used as the first calibration parameter.

[0085] Furthermore, the first calibration parameter may also be obtained by using an ICP algorithm; or by using point-to-plane registration, FGR (fast global registration), RANSAC (random sampling consensus) registration, and the like.

[0086] In the above steps, the determination of the first calibration parameter can automatically realize the external parameter calibration of multiple laser radars, solving the problem of poor consistency of multiple laser radars in complex environments and inaccurate manual calibration.

[0087] Optionally, the candidate calibration parameter determination process further includes steps B1-B4:

[0088] Step B1: extract features of static objects in the first image to obtain image feature data.

[0089] The first image is an image of the vehicle and its surroundings.

[0090] Furthermore, the first image also includes a target. The target is set up because the target has a simple shape and can provide clear and accurate feature points to improve the accuracy of calibration.

[0091] Specifically, the first image is obtained by taking an image of the vehicle and its surroundings by a photographing device, and the texture and edge information of the static object in the first image are extracted to obtain image feature data.

[0092] Furthermore, due to the influence of noise features, an adaptive feature filtering module is used before feature extraction to identify and filter out the influence of noise, so as to improve matching accuracy.

[0093] Step B2: cross-modally match the first point cloud data with the image feature data to obtain static feature point pairs.

[0094] Specifically, the corner point and edge feature information of the static object is obtained from the first point cloud data to obtain point cloud feature data, and the obtained feature data is matched with the image feature data to obtain static feature point pairs.

[0095] Step B3: construct the first equation based on the static feature point pairs.

[0096] The first equation is used to characterize the conversion relationship between the feature points in the first point cloud data and the feature points in the first image data.

[0097] Specifically, the point cloud feature data is used as the dependent variable of the first equation, and the image feature data is used as the independent variable to construct the first equation.

[0098] Step B4: Solve the first equation to obtain the second calibration parameter.

[0099] Specifically, an initial feature parameter is initialized and brought into the first equation for calculation to obtain reference static feature data. The dependent variable feature data in the static feature pair is compared with the reference static feature data to obtain a feature change error, and the initial feature parameter is updated according to the feature change error until the obtained feature change error meets the preset requirements, and the corresponding initial feature parameter is used as the fifth calibration parameter.

[0100] In the above steps, the second calibration parameter determination process can identify feature points in natural scenes, and perform multimodal feature matching in combination with targets, reducing dependence on artificial targets, thereby ensuring the adaptability and accuracy of the second calibration parameters, so that the second calibration parameters can effectively meet the calibration requirements of different environments.

[0101] Optionally, before extracting features of the static object in the first image to obtain image feature data, steps C1-C4 are included:

[0102] Step C1, determining the second image.

[0103] The second image is an image of the reference object that has been blurred and filtered.

[0104] The reference object is a pre-placed object used to perform internal parameter correction on the shooting device. Exemplarily, the reference object may be a chessboard picture.

[0105] Specifically, an image of a reference object is captured by a capturing device, and blur filtering is performed on the acquired image to obtain a second image.

[0106] Furthermore, blur filtering can be performed by filtering the image of the reference object based on the Laplacian operator to remove blurry images.

[0107] Among them, the Laplacian operator first converts the image of the reference object into a grayscale image, then the grayscale image of a single channel is calculated through the Laplacian 3x3 convolution kernel to obtain a response map, and finally the variance of the response map is calculated. For the checkerboard image, the variance is compared with the threshold set empirically, and the image judged to be blurred is discarded to obtain the second image.

[0108] Furthermore, the Laplacian operator can be expressed as follows:

[0109] Step C2: perform corner point detection on the second image to obtain first coordinate information.

[0110] The first coordinate information is the pixel coordinates of the corner points in the reference object under distortion.

[0111] Specifically, corner point detection is performed on the reference object in the second image to obtain first coordinate information of distorted corner points.

[0112] Furthermore, corner detection may use an image feature detection method, such as SURF (Speeded Up Robust Features) or ORB (Orbital Rotation Invariant Features) algorithm, to further enhance the effect of image quality screening.

[0113] Step C3: determining the distortion coefficient according to the first coordinate information and the second coordinate information.

[0114] The second coordinate information is the pixel coordinates of the corner points in the reference object without distortion.

[0115] Specifically, a distortion equation is constructed according to the first coordinate information and the second coordinate information, and the distortion equation is solved by a least square method to obtain a distortion coefficient.

[0116] Step C4: determine the target internal and external parameters based on the distortion coefficient and the reference internal and external parameters.

[0117] The reference internal and external parameters are the internal and external parameters of the shooting device without distortion.

[0118] Specifically, the reference internal and external parameters are adjusted according to the distortion coefficient according to the Levenberg–Marquardt algorithm nonlinear optimization algorithm until the error is minimized, and the internal and external parameters with the minimum error are used as the target internal and external parameters.

[0119] Furthermore, the acquisition of the target internal and external parameters can also adopt the Laplacian operator, or use other blur detection algorithms, such as the Sobel operator, Canny edge detection, or classify and judge the image quality through a machine learning model (such as a convolutional neural network).

[0120] For example, Figure 4 As shown, when the camera internal and external calibration button is selected, the target internal and external parameters of the shooting device, that is, the camera, are determined in the above manner, and the shooting device is calibrated according to the obtained target internal and external parameters.

[0121] Furthermore, when the target internal and external parameters are obtained, they will be displayed on the display interface, such as Figure 8As shown, the obtained camera internal parameters are fx=433.56, fy=433.24, cx=691.88, cy=400.14, k1=-0.0814, k2=0.0050, p1=-0.0001, p2=0.0009.

[0122] Optionally, the candidate calibration parameter determination process further includes steps D1-D4:

[0123] Step D1: extract features of dynamic objects and static objects in the third point cloud data respectively to obtain dynamic feature data and static feature data.

[0124] Among them, the third point cloud data is collected by the second point cloud sensor.

[0125] The second point cloud sensor is a point cloud acquisition sensor that is of a different type from the reference point cloud sensor and is located in the same common viewing area.

[0126] Exemplarily, the second point cloud sensor may be a millimeter wave radar.

[0127] The dynamic feature data is used to represent the relative position change information of the dynamic object, and the static feature data is used to represent the position information of the static object.

[0128] Specifically, features capable of characterizing the position transformation of dynamic objects are extracted from the third point cloud data as dynamic feature data; and position information of static objects in the third point cloud data is used as static feature data.

[0129] Step D2: construct a second equation based on the dynamic feature data and the first point cloud data.

[0130] The second equation is used to characterize the relationship between motion data and time when a dynamic object is running.

[0131] Specifically, the second equation is constructed according to the relationship between the dynamic feature data, the first point cloud data and time.

[0132] Step D3: construct a third program based on the static feature data and the first point cloud data.

[0133] Among them, the third equation represents the relationship between the position and posture of a static object.

[0134] Specifically, the position information corresponding to the static object is matched from the first point cloud data, and a third equation is constructed based on the acquired information and the static feature data.

[0135] Exemplarily, assuming that the coordinates of the static object in the reference point cloud sensor are A, and the coordinates of the static object in the second point cloud sensor are B, then the third equation is A=TB.

[0136] Step D4: determining a third calibration parameter according to the second equation and the third equation.

[0137] Specifically, the first equation and the second equation are solved respectively, and the two reference third calibration parameters obtained by the solutions are jointly corrected to obtain the third calibration parameters.

[0138] Furthermore, the third calibration parameters can also be obtained by dynamically aligning multiple targets by introducing a multi-target tracking algorithm (such as Kalman filtering, multi-hypothesis tracking MHT), or by combining background modeling and feature tracking to perform consistency analysis on dynamic objects in the scene, so that the obtained third calibration parameters have higher robustness and accuracy.

[0139] In the above steps, the determination of the third calibration parameter improves the accuracy and robustness of the second point cloud sensor calibration and solves the problem of insufficient accuracy of a single feature matching method.

[0140] Optionally, the candidate calibration parameter determination process further includes steps E1-E4:

[0141] Step E1: pre-integrate the motion data to obtain relative posture transformation data.

[0142] The motion data is data collected by the motion sensor.

[0143] Among them, pre-integration is used to determine the relative posture transformation data between two key frames.

[0144] Specifically, the angular velocity and acceleration corresponding to every two adjacent key frames of the motion data are pre-integrated to obtain relative posture transformation data.

[0145] Step E2: determining posture change data according to the relative motion change between adjacent frames in the first point cloud data.

[0146] Specifically, the posture change data posture transformation information is determined according to the change in the relative position and posture of the dynamic object between each two adjacent frames in the first point cloud data.

[0147] Step E3: determining the posture change equation according to the relative posture transformation data and the posture change data.

[0148] Specifically, the relative posture transformation data and the posture change data are matched according to the timestamps, and the redundant data are eliminated. The aligned relative posture transformation data are used as the dependent variable, and the aligned posture change data are used as the independent variable to construct the posture change equation.

[0149] Step E4: Solve the posture change equation to obtain the fourth calibration parameter.

[0150] Specifically, a posture transformation parameter is initialized and brought into the posture transformation equation to calculate and obtain the reference phase posture transformation. The relative posture transformation data is compared with the reference phase posture transformation to obtain the posture change error, and the posture transformation parameter is updated according to the posture change error until the posture change error meets the preset requirements, and the corresponding posture transformation parameter is used as the fourth calibration parameter.

[0151] Furthermore, the fourth calibration parameter is also obtained through a method of time synchronization or sensor fusion filtering, such as an algorithm based on timestamp alignment and data interpolation.

[0152] Furthermore, a joint optimization algorithm (such as factor graph optimization) can be used to achieve the registration of IMU and LiDAR data. The angular velocity and acceleration data of the IMU can also be directly matched with the LiDAR point cloud.

[0153] In the above step, the determination of the fourth calibration parameter can improve the alignment accuracy between the reference point cloud sensor and the motion sensor, ensure the consistency of the posture data, and effectively solve the posture error problem caused by trajectory deviation.

[0154] Optionally, the candidate calibration parameter determination process further includes steps F1-F3:

[0155] Step F1 determines the relative change data of the dynamic object between different frames according to the first point cloud data.

[0156] Specifically, the position transformation of the dynamic object between different frames is determined according to the first point cloud data to obtain relative change data.

[0157] For example, the relative position of object 1 changes when it moves from point A to point B.

[0158] Step F2: perform trajectory association based on the relative change data and the positioning sequence data to obtain a trajectory association equation.

[0159] Among them, the positioning sequence data is used to represent the coordinate information of the vehicle and its surrounding environment.

[0160] Specifically, the trajectory association is performed based on the relative change data and the positioning sequence data to obtain the corresponding position information under each action, and the corresponding relative change data is used as the dependent variable; the aligned positioning sequence data is used as the independent variable to construct the trajectory association equation.

[0161] Step F3, solving the trajectory correlation equation to obtain the fifth calibration parameter.

[0162] Specifically, a trajectory association parameter is initialized and brought into the trajectory association equation to calculate and obtain reference relative change data. The relative change data is compared with the reference relative change data to obtain a trajectory change error, and the initial trajectory association parameter is updated according to the trajectory change error until the obtained trajectory change error meets the preset requirements, and the corresponding trajectory association parameter is used as the fifth calibration parameter.

[0163] Furthermore, the fifth calibration parameter can also be obtained by using a method combining trajectory alignment and error optimization.

[0164] Among them, the error optimization method can use robust estimation or Bayesian filtering to enhance the calibration accuracy and stability.

[0165] For example, in an environment with poor RTK signals, motion data can be used as a compensation data source, combined with motion data to form a "laser-IMU-RTK" fusion positioning method, or the trajectory can be corrected through historical trajectory data matching, map alignment, etc., to achieve high-precision posture alignment.

[0166] In the above step, the determination of the fifth calibration parameter realizes high-precision alignment of the reference point cloud sensor and the positioning sensor in multiple environments, ensuring the robustness of the system under conditions of signal quality fluctuations.

[0167] The technical solution of this embodiment transforms the first coordinate system according to the first transformation relationship to obtain the second coordinate system, which provides a transformation basis for the construction of candidate calibration parameters; constructs candidate calibration parameters according to the transformation relationship between the second coordinate system and the coordinate system of at least one target sensor, and the construction of candidate calibration parameters can provide different calibration parameters for different sensors, laying the foundation for sensor calibration; matches target calibration parameters from candidate calibration parameters according to the target sensor type; calibrates the corresponding target sensor according to the target calibration parameters, and can transform different sensors from their respective coordinate systems to the same coordinate system, solving the external parameter and timing synchronization problems between different sensors, and improving the accuracy of data fusion. This method effectively solves the problems of errors, poor adaptability to complex scenes, and insufficient accuracy in the calibration process of multiple sensors in autonomous driving vehicles by jointly calibrating multiple sensors, and meets the requirements of autonomous driving systems for high-precision multi-sensor fusion.

[0168] Fig. 9 This is a schematic diagram of the structure of a multi-sensor calibration device provided by an embodiment of the present invention. This embodiment is applicable to the case of joint calibration of multiple sensors. The multi-sensor calibration device can be implemented in the form of hardware and / or software. The multi-sensor calibration device can be configured in any electronic device with network communication function. Fig. 9As shown, the device includes: a second coordinate system determination module 210, a candidate calibration parameter determination module 220, a target calibration parameter determination module 230 and a calibration module 240, wherein:

[0169] The second coordinate system determination module 210 is used to transform the first coordinate system according to the first transformation relationship to obtain the second coordinate system; the first transformation relationship is used to characterize the transformation relationship between the first coordinate system and the reference coordinate system; the first coordinate system is the coordinate system of the reference point cloud sensor; the reference coordinate system is a coordinate system established with the preset center of the vehicle as the coordinate origin;

[0170] Candidate calibration parameter determination module 220: used to construct candidate calibration parameters according to the transformation relationship between the second coordinate system and the coordinate system of at least one target sensor; the candidate calibration parameters are used to characterize the transformation relationship between the second coordinate system and the coordinate system of each sensor;

[0171] The target calibration parameter determination module 230 is used to match the target calibration parameters from the candidate calibration parameters according to the target sensor type;

[0172] The calibration module 240 is used to calibrate the corresponding target sensor according to the target calibration parameters.

[0173] Optionally, the candidate calibration parameter determination module 220 includes:

[0174] Point cloud data determination unit: used to determine first point cloud data and second point cloud data; the first point cloud data is data collected by the reference point cloud sensor in the second coordinate system; the second point cloud data is point cloud data of the vehicle and the surrounding environment collected by the first point cloud sensor; the first point cloud sensor is a point cloud collection sensor of the same type as the reference point cloud sensor and located in the same common view area;

[0175] A reference voxel unit determining unit: used for voxelizing the first point cloud data to obtain a reference voxel unit;

[0176] Initial transformation matrix determination unit: used to determine the reference density function according to the reference voxel unit and generate the initial transformation matrix; the reference density function is used to characterize the distribution of the point cloud in the reference voxel unit; the initial change matrix is ​​obtained by rotation and translation;

[0177] A target density function determination unit is used to map the second point cloud data to the reference voxel unit according to the initial transformation matrix and determine the target density function; the target density function is used to characterize the distribution of each point cloud in the second point cloud data within the reference voxel unit;

[0178] The first calibration parameter determination unit is used to determine the first calibration parameter according to the reference density function and the target density function.

[0179] Optionally, the candidate calibration parameter determination module 220 further includes:

[0180] An image feature data determination unit is used to extract features of static objects in a first image to obtain image feature data; the first image is an image of the vehicle and its surroundings;

[0181] A static feature point pair determination unit: used for performing cross-modal matching between the first point cloud data and the image feature data to obtain a static feature point pair;

[0182] The first equation will construct a unit for constructing the first equation according to the static feature point pair; the first equation is used to characterize the conversion relationship between the feature points in the first point cloud data and the feature points in the first image data;

[0183] The second calibration parameter determination unit is used to solve the first equation to obtain the second calibration parameter.

[0184] Optionally, the candidate calibration parameter determination module 220 includes:

[0185] A second image determination unit: used to determine a second image; the second image is an image of a reference object that has been blurred and filtered; the reference object is an object that is placed in advance and used to perform internal parameter correction on the shooting device;

[0186] A first coordinate information determining unit: used to detect corner points of the second image to obtain first coordinate information; the first coordinate information is the pixel coordinates of the corner points in the reference object under distortion;

[0187] A distortion coefficient determination unit: used to determine the distortion coefficient according to the first coordinate information and the second coordinate information; the second coordinate information is the pixel coordinates of the corner point in the reference object without distortion;

[0188] Target internal and external parameter determination unit: used to determine target internal and external parameters based on distortion coefficient and reference internal and external parameters; reference internal and external parameters are internal and external parameters of the shooting device in the absence of distortion.

[0189] Optionally, the candidate calibration parameter determination module 220 further includes:

[0190] Feature data determination unit: used to extract features of dynamic objects and static objects in the third point cloud data respectively to obtain dynamic feature data and static feature data; the third point cloud data is collected by the second point cloud sensor; the second point cloud sensor is a point cloud collection sensor of a different type from the reference point cloud sensor and located in the same common view area

[0191] A second equation determination unit: used to construct a second equation according to the dynamic feature data and the first point cloud data; the second equation is used to characterize the relationship between the motion data and time when the dynamic object is running;

[0192] A third equation determination unit: used for constructing a third equation according to the static feature data and the first point cloud data; the third equation represents the relationship between the position and posture of the static object;

[0193] The third calibration parameter determination unit is used to determine the third calibration parameter according to the second equation and the third equation.

[0194] Optionally, the candidate calibration parameter determination module 220 further includes:

[0195] Relative posture transformation data determination unit: used for pre-integrating motion data to obtain relative posture transformation data; the motion data is data collected by the motion sensor; the pre-integration is used to determine the relative posture transformation data between two key frames;

[0196] A posture change data determining unit: used to determine the posture change data according to the relative motion change between adjacent frames in the first point cloud data;

[0197] An attitude change equation determination unit: used for determining an attitude change equation according to relative posture transformation data and attitude change data;

[0198] The fourth calibration parameter determination unit is used to solve the posture change equation to obtain the fourth calibration parameter.

[0199] Optionally, the candidate calibration parameter determination module 220 further includes:

[0200] A relative change determination unit: used to determine relative change data of a dynamic object between different frames according to the first point cloud data;

[0201] Track association equation determination unit: used to associate tracks with positioning sequence data based on relative change data to obtain track association equations; positioning sequence data is used to characterize coordinate information of the vehicle and the surrounding environment;

[0202] The fifth calibration parameter determination unit is used to solve the trajectory association equation to obtain the fifth calibration parameter.

[0203] The multi-sensor calibration device provided in the embodiment of the present invention can execute the multi-sensor calibration method provided in any embodiment of the present invention mentioned above, and has the corresponding functions and beneficial effects of executing the multi-sensor calibration method. For detailed process, please refer to the relevant operations of the multi-sensor calibration method in the aforementioned embodiment.

[0204] Fig.10A schematic diagram of the structure of an electronic device for implementing the multi-sensor calibration method of an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0205] like Fig.10 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0206] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0207] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a multi-sensor calibration method.

[0208] In some embodiments, the multi-sensor calibration method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the multi-sensor calibration method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the multi-sensor calibration method in any other appropriate manner (e.g., by means of firmware).

[0209] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0210] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0211] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0212] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0213] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0214] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0215] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0216] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-sensor calibration method, characterized in that: include: The first coordinate system is transformed according to a first transformation relationship to obtain a second coordinate system; the first transformation relationship is used to characterize the transformation relationship between the first coordinate system and the reference coordinate system; The first coordinate system is a coordinate system of a reference point cloud sensor; the reference coordinate system is a coordinate system established with a preset center of the vehicle as a coordinate origin; Constructing candidate calibration parameters according to a transformation relationship between the second coordinate system and a coordinate system of at least one target sensor; the candidate calibration parameters are used to characterize the transformation relationship between the second coordinate system and the coordinate systems of each sensor; matching target calibration parameters from candidate calibration parameters according to the target sensor type; The corresponding target sensor is calibrated according to the target calibration parameters.

2. The method according to claim 1, characterized in that The candidate calibration parameter determination process includes: Determine first point cloud data and second point cloud data; the first point cloud data is data collected by a reference point cloud sensor in a second coordinate system; the second point cloud data is point cloud data of the vehicle and the surrounding environment collected by the first point cloud sensor; the first point cloud sensor is a point cloud collection sensor of the same type as the reference point cloud sensor and located in the same common view area; voxelize the first point cloud data to obtain a reference voxel unit; Determine a reference density function according to the reference voxel unit and generate an initial transformation matrix; the reference density function is used to characterize the distribution of the point cloud in the reference voxel unit; the initial transformation matrix is ​​obtained by rotation and translation; Mapping the second point cloud data to a reference voxel unit according to the initial transformation matrix, and determining a target density function; the target density function is used to characterize the distribution of each point cloud in the second point cloud data within the reference voxel unit; A first calibration parameter is determined according to the reference density function and the target density function.

3. The method according to claim 1, characterized in that The candidate calibration parameter determination process further includes: Extracting features of static objects in a first image to obtain image feature data; the first image is an image of the vehicle and its surroundings; Cross-modally matching the first point cloud data with the image feature data to obtain static feature point pairs; Constructing a first equation based on the static feature point pair; the first equation is used to characterize the conversion relationship between the feature points in the first point cloud data and the feature points in the first image data; The first equation is solved to obtain a second calibration parameter.

4. The method according to claim 3, characterized in that Before extracting features of the static object in the first image to obtain image feature data, the method includes: Determine a second image; the second image is an image of a reference object that has been blurred and filtered; the reference object is an object that is placed in advance and used to perform internal parameter correction on the shooting device; Performing corner point detection on the second image to obtain first coordinate information; the first coordinate information is the pixel coordinates of the corner points in the reference object under distortion; Determine the distortion coefficient according to the first coordinate information and the second coordinate information; the second coordinate information is the pixel coordinate of the corner point in the reference object without distortion; The target internal and external parameters are determined based on the distortion coefficient and the reference internal and external parameters; the reference internal and external parameters are the internal and external parameters of the shooting device in a distortion-free state.

5. The method according to claim 1, characterized in that The candidate calibration parameter determination process further includes: Extracting features of dynamic objects and static objects in the third point cloud data respectively to obtain dynamic feature data and static feature data; the third point cloud data is collected by a second point cloud sensor; the second point cloud sensor is a point cloud collection sensor of a different type from the reference point cloud sensor and located in the same common viewing area; Constructing a second equation according to the dynamic feature data and the first point cloud data; the second equation is used to characterize the relationship between motion data and time when the dynamic object is running; Constructing a third equation based on the static feature data and the first point cloud data; the third equation represents the relationship between the position and posture of the static object; A third calibration parameter is determined according to the second equation and the third equation.

6. The method according to claim 1, characterized in that The candidate calibration parameter determination process further includes: Pre-integrating the motion data to obtain relative posture transformation data; the motion data is data collected by the motion sensor; the pre-integration is used to determine the relative posture transformation data between two key frames; Determining posture change data according to the relative motion change between adjacent frames in the first point cloud data; Determine a posture change equation according to the relative posture transformation data and the posture change data; The posture change equation is solved to obtain a fourth calibration parameter.

7. The method according to claim 1, characterized in that The candidate calibration parameter determination process further includes: Determine relative change data of the dynamic object between different frames according to the first point cloud data; Performing trajectory association based on the relative change data and the positioning sequence data to obtain a trajectory association equation; the positioning sequence data is used to characterize the coordinate information of the vehicle and the surrounding environment; The trajectory association equation is solved to obtain a fifth calibration parameter.

8. A multi-sensor calibration device, characterized in that: include: A second coordinate system determination module, used to transform the first coordinate system according to the first transformation relationship to obtain a second coordinate system; The first conversion relationship is used to characterize the conversion relationship between the first coordinate system and the reference coordinate system; the first coordinate system is the coordinate system of the reference point cloud sensor; the reference coordinate system is a coordinate system established with the preset center of the vehicle as the coordinate origin; a candidate calibration parameter determination module, configured to construct candidate calibration parameters according to a transformation relationship between the second coordinate system and a coordinate system of at least one target sensor; the candidate calibration parameters are used to characterize a transformation relationship between the second coordinate system and the coordinate systems of each sensor; a target calibration parameter determination module, for matching target calibration parameters from candidate calibration parameters according to the target sensor type; The calibration module is used to calibrate the corresponding target sensor according to the target calibration parameters.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the multi-sensor calibration method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the multi-sensor calibration method according to any one of claims 1 to 7 when executed.

Images