CN114080547A

CN114080547A - Calibration method and calibration device for multiple groups of laser radar external parameters and computer storage medium

Info

Publication number: CN114080547A
Application number: CN202080006493.3A
Authority: CN
Inventors: 宫正
Original assignee: SZ DJI Technology Co Ltd
Current assignee: SZ DJI Technology Co Ltd
Priority date: 2020-06-15
Filing date: 2020-06-15
Publication date: 2022-02-22
Also published as: WO2021253193A1

Abstract

A calibration method, a calibration device and a computer storage medium for multiple groups of laser radar external parameters are provided, wherein the calibration method comprises the following steps: acquiring multiple frames of first point cloud data and multiple frames of second point cloud data which are respectively acquired by a first laser radar and a second laser radar in a scanning calibration scene (S301); constructing a map according to the multi-frame first point cloud data and estimating a motion track of the movable platform relative to a calibration scene (S302); sequentially rotating and translating the multi-frame second point cloud data into a map based on the motion track and the initial values of the external parameters so as to solve multiple groups of estimated external parameters and matching scores corresponding to each group of estimated external parameters (S303); a target outlier is determined based on the estimated outliers of at least a subset of the plurality of sets of estimated outliers (S304), wherein a matching score of each of the at least a subset of the sets of estimated outliers is greater than a threshold score. The calibration external parameters of the multiple groups of laser radars obtained by the method have high precision.

Description

Calibration method and calibration device for multiple groups of laser radar external parameters and computer storage medium

Description

Technical Field

The present invention generally relates to the field of laser radar technology, and more particularly, to a calibration method, a calibration apparatus, and a computer storage medium for multiple sets of laser radar external parameters.

Background

With the wide application of the vehicle-mounted sensor, the laser radar plays an important role in a vehicle-mounted sensor system due to the high resolution, 360-degree omnibearing and three-dimensional environment sensing capability. In an existing vehicle-mounted sensor system, a plurality of laser radars are generally configured at different positions of a vehicle to achieve non-blind area coverage of point cloud, and meanwhile, the same target object is observed through the plurality of laser radars, so that information such as the shape, the outline, the behavior and the posture of the target object is captured more comprehensively. However, in order to obtain the above advantages, the calibration of the laser radar is an important prerequisite, and the calibration of the laser radar needs to be completed by obtaining the accurate relative poses of a plurality of laser radars, so that the point cloud data of all the laser radars can be accurately unified to the same coordinate system.

At present, the mode of calibrating the external parameters of a plurality of laser radars by using a multi-purpose fixed turntable needs to fix a calibration object, and has high requirements on fields and working environments. And existing rotational registration methods may have a failure problem for small field of view (FOV) lidar. And the measuring points are fewer, and the result precision can not be ensured.

Disclosure of Invention

The present invention has been made to solve at least one of the above problems. Specifically, one aspect of the present application provides a calibration method for multiple sets of lidar parameters, where the multiple sets of lidar parameters include a first lidar and a second lidar, and the first lidar and the second lidar are mounted at different positions of a movable platform, and the calibration method includes:

acquiring multiple frames of first point cloud data and multiple frames of second point cloud data which are respectively collected by the first laser radar and the second laser radar in a scanning calibration scene;

constructing a map according to the multiple frames of the first point cloud data and estimating a motion track of the movable platform relative to the calibration scene;

sequentially rotating and translating the multiple frames of second point cloud data into the map based on the motion tracks and the initial values of the external parameters so as to solve multiple groups of estimated external parameters and matching scores corresponding to each group of estimated external parameters;

determining a target outlier based on estimated outliers of at least a subset of the plurality of sets of estimated outliers, wherein a match score of each of the at least a subset of the sets of estimated outliers is greater than a threshold score.

In another aspect, the present invention provides a calibration apparatus for multiple sets of external parameters of a laser radar, the apparatus comprising:

a movable platform;

the laser radar system comprises a movable platform body, a plurality of groups of laser radars, a plurality of laser radar sensors and a control module, wherein the plurality of groups of laser radars comprise a first laser radar and a second laser radar which are arranged at different positions of the movable platform body;

a memory having stored thereon a computer program for execution by the processor, and a processor for:

Yet another aspect of the present invention provides a computer storage medium, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the aforementioned calibration method for multiple sets of lidar external parameters.

According to the calibration method and the calibration device, the point cloud around all the radar acquisition devices is used, one reference radar is selected to estimate the motion track, meanwhile, a map is built for the surrounding environment, and then the matching scores corresponding to a plurality of groups of estimation external parameters and each group of estimation external parameters are solved. And the matching score is used as registration failure judgment aiming at the problem of laser radar registration failure, and the estimation accuracy is improved by filtering the estimation external parameter of which the matching score is lower than the threshold score, so that a more accurate calibration result is obtained.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive labor.

FIG. 1 shows a schematic diagram of an application scenario in one embodiment of the invention;

FIG. 2 illustrates a schematic diagram of a calibration scenario in one embodiment of the invention;

FIG. 3 is a flow chart illustrating a method for calibrating multiple sets of lidar external parameters in an embodiment of the invention;

FIG. 4 is a flow chart illustrating a method for calibrating multiple sets of lidar external parameters in another embodiment of the present invention;

FIG. 5 illustrates a relationship diagram of weights and covariances in one embodiment of the invention;

FIG. 6 is a block diagram of a calibration apparatus for multiple sets of lidar external parameters in an embodiment of the invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, exemplary embodiments according to the present application will be described in detail below with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.

In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present application. It will be apparent, however, to one skilled in the art, that the present application may be practiced without one or more of these specific details. In other instances, well-known features of the art have not been described in order to avoid obscuring the present application.

It is to be understood that the present application is capable of implementation in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the application to those skilled in the art.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. As used herein, the term "and/or" includes any and all combinations of the associated listed items.

In order to provide a thorough understanding of the present invention, a detailed structure will be set forth in the following description in order to explain the present invention. Alternative embodiments of the invention are described in detail below, however, the invention may be practiced in other embodiments that depart from these specific details.

It should be noted that the laser radar calibration method provided in the embodiment of the present application may be applied not only to an automatic driving scenario but also to a robot navigation scenario, for example, and the embodiment of the present application does not limit a specific application scenario.

In an example, the calibration method for multiple sets of lidar external parameters provided by the embodiment may be applied to an application environment as shown in fig. 1. As shown in fig. 1, the first laser radar 10 and the second laser radar 20 may be installed at any position of a movable platform, such as a vehicle, for example, the first laser radar 10 is installed at the left side of the movable platform, and the second laser radar 20 is installed at the right side of the movable platform, or alternatively, the first laser radar 10 may be installed at the front of the movable platform, and the second laser radar may be installed at the rear of the movable platform, or any other suitable position may be used, and the calibration external parameter of the laser radar may be determined by obtaining the relative position information between the first laser radar 10 and the second laser radar 20 through a calibration method, and the point cloud data collected by the first laser radar 10 and the second laser radar 20 is put into a coordinate system based on the calibration external parameter.

The scan fields of view of the different lidar may or may not overlap. And when the scanning fields of different laser radars have overlap, the overlapping rate is not limited.

The vehicle can comprise two laser radars or a plurality of laser radars, and the vehicle can be an automatic driving vehicle or a common vehicle.

The first lidar 10 and the second lidar 20 are respectively arranged on two sides of the movable platform, and the visual areas have no overlapping parts. At present, the mode of calibrating the external parameters of a plurality of laser radars by using a multi-purpose fixed turntable needs to fix a calibration object, and has high requirements on fields and working environments. And existing rotational registration methods may have a failure problem for small field of view (FOV) lidar. Under the condition, the point cloud images of the two laser radars cannot be directly matched manually or automatically, external parameters need to be measured through measurement or other modes, the number of measurement points is small, the process is complex, and the precision is extremely low.

In view of the above problems, the present invention provides a calibration method for multiple sets of lidar parameters, where the multiple sets of lidar parameters include a first lidar and a second lidar, and the first lidar and the second lidar are mounted at different positions of a movable platform, and the calibration method includes: acquiring multiple frames of first point cloud data and multiple frames of second point cloud data which are respectively collected by the first laser radar and the second laser radar in a scanning calibration scene; constructing a map according to the multi-frame first point cloud data and estimating a motion track of the movable platform relative to the calibration scene; sequentially rotating and translating the multiple frames of second point cloud data into the map based on the motion tracks and the initial values of the external parameters so as to solve multiple groups of estimated external parameters and matching scores corresponding to each group of estimated external parameters; determining a target outlier based on estimated outliers of at least a subset of the plurality of sets of estimated outliers, wherein a match score of each of the at least a subset of the sets of estimated outliers is greater than a threshold score. According to the method, the point clouds around all radar acquisition devices are used, one reference radar is selected to estimate the motion track, a map is built for the surrounding environment, and then the matching scores corresponding to a plurality of groups of estimation external parameters and each group of estimation external parameters are solved. And the matching score is used as registration failure judgment aiming at the problem of laser radar registration failure, and the estimation accuracy is improved by filtering the estimation external parameter of which the matching score is lower than the threshold score, so that a more accurate calibration result is obtained.

It should be noted that, in the calibration method for multiple sets of external parameters of a laser radar provided in the embodiment of the present application, an execution main body may be a calibration device for external parameters of a laser radar, and the device may be implemented as part or all of a computer device for laser radar calibration in a software, hardware, or a combination of software and hardware.

The following describes in detail a calibration method and a calibration apparatus for multiple sets of lidar external references in accordance with the present application with reference to the accompanying drawings. The features of the following examples and embodiments may be combined with each other without conflict.

As an example, as shown in fig. 3, the method for calibrating multiple sets of external parameters of laser radar according to the embodiment of the present invention includes the following steps S301 to S304.

First, as shown in fig. 3, in step S301, multiple frames of first point cloud data PnA and multiple frames of second point cloud data PnB respectively collected in a scanning calibration scene of the first laser radar and the second laser radar are obtained.

In one example, the calibration is performed in a calibration environment as shown in fig. 2, the first lidar 10 and the second lidar 20 are fixed to opposite sides of a movable platform, e.g., the movable platform is a vehicle, the first lidar 10 and the second lidar 20 are fixed to left and right sides of the movable platform, and the calibration scenario may be any suitable indoor or outdoor fixed location, and the calibration scenario includes rich, static structural features, such as walls, shafts, geometries, etc. In the embodiment of the present invention, multiple lidar may be rigidly mounted at any position on the movable platform body, for example, the first lidar and the second lidar scan the calibration scene while the movable platform moves according to a predetermined motion trajectory in the calibration scene, and simultaneously collect multi-frame point cloud data of each lidar. The requirement on the motion trajectory is that after the movable platform moves according to a predetermined motion trajectory in the calibration scene, an overlapping rate of a scene range scanned by the first laser radar and a scene range scanned by the second laser radar reaches a preset threshold, for example, the preset threshold is 80%. The motion trajectory is a non-linear motion, for example, the motion trajectory may be a U-shaped motion, a circular motion, an elliptical motion, a square motion, a semicircular motion, or the like, and in another example, the movable platform is fixed, and the first laser radar and the second laser radar scan the calibration scene while the calibration scene moves relative to the movable platform, and collect multi-frame point cloud data of each laser radar at the same time, in this case, the calibration scene may also move according to the motion trajectory.

Herein, the first point cloud data is referenced to a coordinate system of the first lidar and the second point cloud data is referenced to a coordinate system of the second lidar.

In this context, the external reference is a calibration matrix that converts the coordinate system of the second lidar to the coordinate system of the first lidar, the calibration matrix comprising 3 rotation parameters and 3 translation parameters (e.g. x, y, z), 3 rotation parameters (pitch, roll, yaw).

The field of view of the first lidar 10 and the second lidar 20 may be any suitable field of view, for example the first lidar 10 and the second lidar 20 each have a smaller field of view, in one example the field of view of the first lidar 10 (e.g. the horizontal field of view) is less than 100 °, e.g. 60 °, 70 °, 80 °, 90 °; and/or the field of view of the second lidar 20 (e.g. the horizontal field of view) is less than 100 °, e.g. 60 °, 70 °, 80 °, 90 °, etc. The fields of view of first lidar 10 and second lidar 20 may not overlap, or may partially overlap.

In step S302, a map is constructed according to the multiple frames of first point cloud data, and a motion trajectory of the movable platform relative to the calibration scene is estimated.

After it is determined that the coincidence rate of the scanning tracks of the first laser radar and the second laser radar in the calibration scene is greater than a preset threshold, the specific numerical value of the preset threshold may be reasonably set according to actual needs, for example, may be 70%, 80%, 90%, and the like, a map (for example, a short-time sub-map) is constructed according to the multiple frames of the first point cloud data, and the movement track of the movable platform relative to the calibration scene is estimated. The coincidence of the scanning trajectories of the first lidar and the second lidar in the calibration scenario may be determined by any suitable method known to those skilled in the art, for example, by the motion trajectory, and when the movable platform moves a closed motion trajectory, the coincidence may be satisfied after the motion is completed.

The map is a map of a calibration scene, which may be a point cloud map or a feature map, and may be constructed in a SLAM manner, a feature point registration manner or an automatic registration manner such as ICP, G-ICP, or the like.

And constructing a map of the scanned calibration scene based on the multi-frame first point cloud data acquired within the time from 0 to n, wherein the first laser radar finishes the preset movement track within the time from 0 to n.

In this embodiment, regard as the benchmark radar with first laser radar 10, second laser radar 20 is for waiting to mark the radar, when movable platform set up a plurality of laser radar, can also be with the coordinate system of the first laser radar 10 of the equal calibration value of other laser radar, perhaps, can also regard as the benchmark radar with a laser radar outside the first laser radar, regard as the radar of waiting to mark with the laser radar outside the benchmark radar, mark it to the coordinate system of benchmark radar.

Next, in step S303, based on the motion trajectory and the initial value of the external parameter, the multiple frames of second point cloud data are sequentially rotated and translated into the map, so as to solve multiple sets of estimated external parameters and matching scores corresponding to each set of estimated external parameters.

The match score response rotationally translates the second point cloud data of the respective frame into the map based on the estimated parameters and the degree of match of the respective features in the map, typically the higher the match score, the higher the degree of match.

In the method, the estimated extrinsic parameters of the corresponding frame are solved based on the second point cloud data of each frame of the multiple frames of second point cloud data in sequence, for example, the estimated extrinsic parameters R and the matching score d are solved for the second point cloud data of the 1 st frame to the nth frame in sequence when the second point cloud data of the n frames are collected within the calibration time.

Taking the case of solving a group of estimated outliers R and matching scores d based on one frame of second point cloud data of multiple frames of second point cloud data as an example, the method for solving the estimated outliers R and the matching scores d is described, and it is conceivable that the estimated outliers R and the matching scores d of any one frame of the multiple frames of second point cloud data may be solved according to the methods.

In one example, the multi-frame second point cloud data includes second point cloud data of a current frame, where the current frame may be any one frame of the multi-frame second point cloud data, the initial value of the extrinsic parameter includes a first initial value, and the multi-frame second point cloud data is sequentially rotated and translated into the map based on a motion track (i.e., a motion track estimated based on the multi-frame first point cloud data), the initial value of the extrinsic parameter, and a synchronous timestamp, so as to solve matching scores corresponding to multiple sets of estimated extrinsic parameters and each set of estimated extrinsic parameters, including the following steps S1 to S3:

first, in step S3, based on the motion trajectory, the first initial value of the extrinsic parameter, and the synchronized timestamp, the second point cloud data of the current frame is rotated and translated into the map to obtain a first converted point cloud of the current frame, where the first initial value of the extrinsic parameter is determined according to a position relationship between the first lidar and the second lidar when the current frame is a first frame of multiple frames, or the first initial value of the extrinsic parameter is an estimated extrinsic parameter with a matching score higher than a threshold score that is solved based on a second point cloud data of a frame before the current frame when the current frame is not the first frame.

Alternatively, the second point cloud data of the current frame may be point cloud data of any one frame of the second point cloud data of the plurality of frames.

Because the motion track is estimated through the first point cloud data, when the second point cloud data of the current frame is rotationally translated to the map, the rotational translation can be performed based on the motion estimation, so that the first point cloud data can be rapidly rotated to the vicinity of the position of the map, for example, the first point cloud data of the current frame is collected as the surface of one object in the map, and the first point cloud data of the current frame can be rapidly rotationally translated to the vicinity of the one object in the map based on the motion track, the first extrinsic initial value and the synchronous timestamp.

When the current frame is a first frame in multiple frames, the first extrinsic parameter initial value is determined according to a position relationship between the first laser radar and the second laser radar, for example, the first extrinsic parameter initial value may be estimated by measuring positions of the first laser radar and the second laser radar and using an engineering drawing, or a feature point may be searched for by machine learning by using a feature matching method, and then a first extrinsic parameter initial value is preliminarily estimated. The optimization efficiency can be improved by setting the initial value of the external parameter.

When the current frame is not the first frame, the first extrinsic parameter initial value is an estimated extrinsic parameter with the matching score higher than a threshold score, which is solved based on second point cloud data of a frame before the current frame. The threshold score can be reasonably set according to actual matching needs, and is not specifically limited, and by means of the setting, estimation external parameters with the matching score lower than the threshold score can be filtered out and are not substituted into next iteration, accuracy of the next iteration is improved, and robustness is improved.

Next, a plurality of neighbor points in the map are obtained, wherein the distance between the neighbor points and the point cloud point in the first conversion point cloud is lower than a threshold distance.

It should be noted that, since each frame of the second point cloud data may include a plurality of point cloud points, the first conversion point cloud corresponding to one frame herein may include a plurality of point cloud points, and the iterative solution may be performed based on one of the point cloud points, or may be performed on at least a part of the point cloud points, and the first conversion point cloud belonging to one frame may substantially correspond to the same external reference.

A plurality of neighbor points in the map that are less than a threshold distance from one point cloud point in the first converted point cloud may be obtained using any suitable method, for example, a plurality of neighbor points PnA that are less than a threshold distance from one point cloud point in the first converted point cloud may be found in the map by a nearest neighbor search algorithm (represented by function NN in equation 1).

In equation 1, P_A ⁿRepresenting neighbor points in a map, M representing a map constructed based on a plurality of frames of first point cloud data, P_B ⁿRepresenting the nth frame of second point cloud data, T, in the n frames of second point cloud data acquired by the second laser radar_guessThe initial value of the external parameter is shown.

Or the environment reconstruction can be completed by automatically matching point clouds through other feature matching algorithms such as LOAM and the like.

As shown in fig. 4, according to the first converted point cloud and the plurality of neighbor points, a current set of estimated outliers and a matching score corresponding to the current frame are determined based on a geometric consistency constraint and a neighbor matching algorithm based on point cloud surface covariance weight matching, when the matching score of the current set of estimated outliers is higher than a threshold score, a next frame of the current frame uses the current set of estimated outliers as the initial outliers, and when the matching score of the current set of estimated outliers is lower than the threshold score, the next frame of the current frame uses the estimated outliers with the matching score higher than the threshold score solved based on a frame of second point cloud data before the current frame as the initial outliers of the next frame. The weighted neighbor matching calculation is carried out in a matching mode based on the covariance weight of the point cloud surface in the process of iteratively solving and estimating the external parameters one frame by one frame, so that the robustness can be effectively improved, the algorithm can simultaneously obtain the matching score, and the matching failure or success judgment is helped through the matching score.

In one example, a current set of estimated outlier and matching score corresponding to the current frame is determined based on a geometry consistency constraint and a neighbor matching algorithm based on covariance weight matching, comprising: determining an optimization equation based on geometric structure consistency constraint and a neighbor matching method based on covariance weight matching; and substituting the first conversion point cloud and the plurality of neighbor points into the optimization equation to solve the current group of estimated external parameters and matching scores corresponding to the current frame.

For example, the optimization equation includes a residual equation, and the current set of estimated outliers and the matching score corresponding to the current frame are obtained by solving the residual equation, where the current set of estimated outliers minimizes an iteration residual of the residual equation, and the smaller the iteration residual, the higher the matching score, for example, the iteration residual includes a first iteration residual and a second iteration residual, where the first iteration residual corresponds to a first matching score and the second iteration residual corresponds to a second matching score, and when the first iteration residual is smaller than the second iteration residual, the first matching score is higher than the second matching score.

Specifically, the optimization equation can be obtained by the following residual equation:

wherein, T is an estimated matching target parameter, C is a characteristic value of a point cloud covariance matrix of a neighbor point adjacent to the corresponding point, d is an iteration residual, and the smaller the iteration residual is, the higher the matching score is.

The estimated external parameter Δ T and the iteration residual d of the current frame can be obtained by solving the optimization equation. The optimization equation can be solved by nonlinear least squares using any suitable optimization method, including but not limited to iterative methods, or there can be a better initial hypothesis (e.g., first initial value of the extrinsic parameter) at the beginning of performing each frame of iteration, and the estimated extrinsic parameter Δ T and the iteration residual d of the current frame can be solved numerically by the Guass-Newton method or the Levenberg-Marquardt method.

The weights of the optimization equation may be set by any suitable method, for example, the weights are eigenvalues of a covariance matrix of a plurality of neighboring points of a point cloud point in the first transformed point cloud, wherein, the number of the plurality of neighbor points can be reasonably set according to actual needs, for example, 10, 20, etc., the plurality of neighbor points includes a first neighbor point and a second neighbor point, the covariance of the first neighbor point is greater than the covariance of the second neighbor point, the weight of the first neighbor point is less than the weight of the second neighbor point, a larger covariance indicates, the noise of the neighboring points of one point cloud point in the first conversion point cloud is more unstable and more unstable, therefore, smaller weights are given, and smaller covariance indicates that the neighbor point of one point cloud point in the first conversion point cloud is a stable feature point, and larger matching weights are given, as shown in fig. 5. By the method, matching accuracy and stability can be improved robustly aiming at radars (namely small FOV) with the field of view lower than a threshold field of view.

The match score is determined by iterating the residuals d, wherein the smaller the iterated residuals, the higher the match score. As shown in fig. 2, after the estimated outlier R and the matching score are obtained by the above-mentioned method, whether the map is successfully matched is determined by the matching score, for example, if the matching score is higher than the threshold score, it indicates that the matching is successful, at this time, the iteration of this time may be saved to obtain the estimated outlier R, if the matching score is smaller than the threshold score, it indicates that the matching is failed, the estimated outlier R is discarded, and the estimated outlier with the matching score higher than the preset score obtained by the previous frame is used as the initial iteration value (i.e., the initial outlier value) of the next frame, or the estimated outlier with the highest matching score before this frame may be used as the initial iteration value of the next frame, so as to improve the optimization efficiency.

And when one iteration is finished, judging whether the current frame is the last frame in the multi-frame second point cloud data, if so, stopping the iteration, otherwise, solving the estimated external parameters and the corresponding matching scores thereof based on the second point cloud data of the next frame according to the method until all the point cloud frames are traversed, thereby solving multiple groups of estimated external parameters.

Further, as shown in fig. 3, in step S304, a target outlier is determined based on the estimated outliers of at least a partial group of the plurality of groups of estimated outliers, wherein a matching score of each of the at least a partial group of estimated outliers is greater than a threshold score.

The final target external parameters can be determined based on the multiple groups of solved estimated external parameters, for example, the average value of the estimated external parameters of at least part of groups is calculated to obtain the target external parameters, and since the estimated external parameters with matching scores lower than the threshold score are filtered out during the iteration, the accurate target external parameters can be obtained by directly using the method of calculating the average value; or fitting the at least partial set of estimated external parameters based on a least squares method to determine the target external parameters; or, the matching score is used as a weight, the estimated external parameters of at least part of groups are fitted based on a weighted least square method to determine the target external parameters, and the higher the matching score is, the better the matching map is, so that the accuracy of the target external parameters is improved by a method of giving weights.

In other examples, a random sampling consistency algorithm may be further used to estimate a final target external parameter (i.e., a parameter matrix) from a plurality of sets of estimated external parameters, so as to improve the accuracy. The six degrees of freedom of the external parameter can be respectively fitted into a straight line through a random sampling consistency algorithm, and the points far away from the straight line are noise points which are filtered, so that the mean value of the points without the noise points is calculated, and the final target external parameter is estimated.

Further, after the target external parameters are determined, the following steps can be carried out: rotationally registering the multi-frame second point cloud data to a coordinate system of the first laser radar based on the target external parameters; performing fusion operation on the rotationally registered multiple frames of second point cloud data and the multiple frames of first point cloud data to obtain point cloud data in a coordinate system of the first laser radar, namely, a global coordinate consistent point cloud Pglobal, and completing final calibration, which can be expressed by the following equation:

P _global＝P _LaserA+T×P _LaserB (3)

the data collected by the plurality of laser radars are calibrated to a coordinate system, so that the point cloud data volume in a scanning scene is increased, and further, the subsequent three-dimensional reconstruction composition and the like are facilitated.

In summary, according to the calibration method of the embodiment of the invention, the point cloud around all the radar acquisition devices is used, one of the reference radars is selected to estimate the motion trajectory, a map is constructed for the surrounding environment, and then the matching scores corresponding to the multiple sets of estimation external parameters and each set of estimation external parameters are solved. And the matching score is used as registration failure judgment aiming at the problem of laser radar registration failure, and the estimation accuracy is improved by filtering the estimation external parameter of which the matching score is lower than the threshold score, so that a more accurate calibration result is obtained, and the robustness is greatly improved. Moreover, the calibration method can realize the calibration of the small FOV laser radar, and the calibration result has high precision.

In the following, a calibration apparatus for multiple sets of external parameters of laser radar according to an embodiment of the present invention is described with reference to fig. 6, where the calibration apparatus can be used to implement the aforementioned calibration method. Wherein the aforementioned features may be incorporated into this embodiment.

As shown in fig. 6, a calibration apparatus 600 according to an embodiment of the present invention includes a movable platform 602, and multiple sets of laser radars, where the multiple sets of laser radars include a first laser radar 603 and a second laser radar 604, and the first laser radar 603 and the second laser radar 604 are installed at different positions of a body of the movable platform. In certain embodiments, movable platform 602 comprises at least one of an unmanned aerial vehicle, an automobile, a remote control car, a robot, a boat.

The field of view of the first and

second lidar

603, 604 may be any suitable field of view, for example the first and

second lidar

603, 604 each have a smaller field of view, in one example the field of view of the first lidar 603 (e.g. horizontal field of view) is less than 100 °, for example 60 °, 70 °, 80 °, 90 °; and/or the field of view of the second lidar 604 (e.g. horizontal field of view) is less than 100 °, e.g. 60 °, 70 °, 80 °, 90 °, etc. The fields of view of first lidar 10 and second lidar 20 may not overlap, or may partially overlap.

In some embodiments, as shown in fig. 6, the calibration apparatus 600 further comprises one or more processors 605, one or more memories 606, the one or more processors 605 working together or separately. Optionally, the calibration device 600 may further include at least one of an input device (not shown), an output device (not shown), and an image sensor (not shown), which are interconnected via a bus system and/or other form of connection mechanism (not shown).

The memory 606 is used for storing thereon a computer program run by the processor, for example, for storing corresponding steps and program instructions for implementing a calibration method for multiple sets of lidar external parameters according to an embodiment of the present invention. May include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, Random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, Read Only Memory (ROM), hard disk, flash memory, etc.

The input device may be a device used by a user to input instructions and may include one or more of a keyboard, a mouse, a microphone, a touch screen, and the like.

The output device may output various information (e.g., images or sounds) to the outside (e.g., a user), and may include one or more of a display, a speaker, and the like, for outputting the point cloud frame as an image or video, or outputting the constructed map as an image or video.

The communication interface (not shown) is used for communication between the calibration apparatus and other devices or communication between the calibration apparatus, including wired or wireless communication. The ranging device may access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G, 5G, or a combination thereof. In one exemplary embodiment, the communication interface receives a broadcast signal or broadcast associated information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, the communication interface further comprises a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

Processor 605 may be a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in calibration arrangement 600 to perform desired functions. The processor is capable of executing program instructions stored in the memory to perform the method for calibrating multiple sets of lidar parameters described herein in embodiments of the invention. For example, a processor can include one or more embedded processors, processor cores, microprocessors, logic circuits, hardware Finite State Machines (FSMs), Digital Signal Processors (DSPs), or a combination thereof. In this embodiment, the processor comprises a Field Programmable Gate Array (FPGA), wherein the operational circuitry of the lidar may be part of the Field Programmable Gate Array (FPGA).

In one example, a memory having stored thereon a computer program for execution by the processor, and a processor to: acquiring multiple frames of first point cloud data and multiple frames of second point cloud data which are respectively collected by the first laser radar and the second laser radar in a scanning calibration scene; constructing a map according to the multiple frames of the first point cloud data and estimating a motion track of the movable platform relative to the calibration scene; sequentially rotating and translating the multiple frames of second point cloud data into the map based on the motion tracks and the initial values of the external parameters so as to solve multiple groups of estimated external parameters and matching scores corresponding to each group of estimated external parameters; determining a target outlier based on estimated outliers of at least a subset of the plurality of sets of estimated outliers, wherein a match score of each of the at least a subset of the sets of estimated outliers is greater than a threshold score.

In one example, the calibration apparatus 600 further comprises a calibration scenario 601 configured to be movable outside the movable platform according to a predetermined motion trajectory, for example the movable platform is fixed, and the calibration scenario is configured to move around the movable platform while the first lidar and the second lidar scan the calibration scenario to acquire point cloud data.

In one example, the multi-frame second point cloud data includes second point cloud data of a current frame, the initial value of the extrinsic parameter includes a first initial value of the extrinsic parameter, and the processor is configured to sequentially rotate and translate the multi-frame second point cloud data into the map based on the motion trajectory and the initial value of the extrinsic parameter to solve matching scores corresponding to a plurality of sets of estimated extrinsic parameters and each set of estimated extrinsic parameters, including: based on the motion track and the first extrinsic parameter initial value, rotationally translating the second point cloud data of the current frame into the map to obtain a first converted point cloud of the current frame, wherein when the current frame is a first frame in the multiple frames, the first extrinsic parameter initial value is determined according to the position relation of the first laser radar and the second laser radar, or when the current frame is not the first frame, the first extrinsic parameter initial value is an estimated extrinsic parameter with a matching score higher than a threshold score, which is solved based on second point cloud data of a frame before the current frame; acquiring a plurality of neighbor points in the map, wherein the distance between the neighbor points and the point cloud point in the first conversion point cloud is less than a threshold distance; according to the first conversion point cloud and the plurality of neighbor points, a current group estimation outlier and a matching score corresponding to the current frame are determined based on a geometric consistency constraint and a neighbor matching algorithm based on covariance weight matching, when the matching score of the current group estimation outlier is higher than a threshold score, the current group estimation outlier is used as the initial outlier value by the next frame of the current frame, and when the matching score of the current group estimation outlier is lower than the threshold score, the estimated outlier with the matching score higher than the threshold score, which is solved based on second point cloud data of a frame before the current frame, is used as the initial outlier value of the next frame by the next frame of the current frame.

In one example, determining, from the first transformed point cloud and the plurality of neighbor points, a current set of estimated outlier and matching scores corresponding to the current frame based on a geometry consistency constraint and a neighbor matching algorithm based on covariance weight matching, comprises: determining an optimization equation based on geometric structure consistency constraint and a neighbor matching method based on covariance weight matching; and substituting the first conversion point cloud and the plurality of neighbor points into the optimization equation to solve the current group of estimated external parameters and matching scores corresponding to the current frame.

In one example, the optimization equation includes a residual equation, and the current set of estimated outliers and the matching score corresponding to the current frame are obtained by solving the residual equation, where the current set of estimated outliers minimizes an iterative residual of the residual equation, and the smaller the iterative residual, the higher the matching score.

In one example, the weight is an eigenvalue of a covariance matrix of the plurality of neighbor points, wherein the plurality of neighbor points includes a first neighbor point and a second neighbor point, and the covariance of the first neighbor point is greater than the covariance of the second neighbor point, then the weight of the first neighbor point is less than the weight of the second neighbor point.

In one example, determining the target external parameters based on the estimated external parameters of at least a partial group of the plurality of groups of estimated external parameters includes: averaging the estimated external parameters of the at least partial group to obtain a target external parameter; or fitting the estimated external parameters of the at least partial group based on a least squares method to determine the target external parameters; or fitting the at least part of the groups of estimated external parameters based on a weighted least squares method by taking the matching scores as weights to determine the target external parameters.

In one example, the movable platform is configured to scan the calibration scene with the first lidar and the second lidar while moving according to a predetermined trajectory in the calibration scene.

In one example, the processor 605 is configured to: and after the coincidence rate of the scanning tracks of the first laser radar and the second laser radar in the calibration scene is determined to be greater than a preset threshold value, constructing a map according to the multiple frames of the first point cloud data and estimating the motion track of the movable platform relative to the calibration scene.

In one example, a coincidence of the scan trajectory of the first lidar and the scan trajectory of the second lidar is above a threshold coincidence.

In one example, the external reference is a calibration matrix that converts the coordinate system of the second lidar to the coordinate system of the first lidar, the calibration matrix comprising 3 rotation parameters and 3 translation parameters.

In one example, the processor is further configured to: rotationally registering the multi-frame second point cloud data to a coordinate system of the first laser radar based on the target external parameters; and performing fusion operation on the rotationally registered multiple frames of second point cloud data and the multiple frames of first point cloud data to obtain point cloud data under the coordinate system of the first laser radar.

According to the calibration device provided by the embodiment of the invention, the point cloud around all the radar acquisition devices is used, one reference radar is selected to estimate the motion track, a map is constructed for the surrounding environment, and then the matching scores corresponding to a plurality of groups of estimation external parameters and each group of estimation external parameters are solved. And the matching score is used as registration failure judgment aiming at the problem of laser radar registration failure, and the estimation accuracy is improved by filtering the estimation external parameter of which the matching score is lower than the threshold score, so that a more accurate calibration result is obtained, and the robustness is greatly improved. Moreover, the calibration method can realize the calibration of the small FOV laser radar, and the calibration result has high precision.

In addition, the embodiment of the invention also provides a computer storage medium, and the computer storage medium is stored with the computer program. One or more computer program instructions may be stored on the computer-readable storage medium, the processor may execute the program instructions stored in the memory to implement the functions (implemented by the processor) of the embodiments of the present invention described herein and/or other desired functions, for example, to execute the corresponding steps of the calibration method for multiple sets of lidar external parameters according to the embodiments of the present invention, and various applications and various data, such as various data used and/or generated by the applications, may also be stored in the computer-readable storage medium.

For example, the computer storage medium may include, for example, a memory card of a smart phone, a storage component of a tablet computer, a hard disk of a personal computer, a Read Only Memory (ROM), an Erasable Programmable Read Only Memory (EPROM), a portable compact disc read only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media. For example, a computer readable storage medium may contain computer readable program code for converting point cloud data into a two-dimensional image, and/or computer readable program code for three-dimensional reconstruction of point cloud data, and the like.

It should be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic Gate circuit for implementing a logic function on a data signal, an asic having a suitable combinational logic Gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), and the like.

Although the illustrative embodiments have been described herein with reference to the accompanying drawings, it is to be understood that the foregoing illustrative embodiments are merely exemplary and are not intended to limit the scope of the invention thereto. Various changes and modifications may be effected therein by one of ordinary skill in the pertinent art without departing from the scope or spirit of the present invention. All such changes and modifications are intended to be included within the scope of the present invention as set forth in the appended claims.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described device embodiments are merely illustrative, and for example, the division of the units is only one logical functional division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another device, or some features may be omitted, or not executed.

In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

Similarly, it should be appreciated that in the description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the invention and aiding in the understanding of one or more of the various inventive aspects. However, the method of the present invention should not be construed to reflect the intent: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.

It will be understood by those skilled in the art that all of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where such features are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.

Furthermore, those skilled in the art will appreciate that while some embodiments described herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the claims, any of the claimed embodiments may be used in any combination.

The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or Digital Signal Processor (DSP) may be used in practice to implement some or all of the functionality of some of the modules according to embodiments of the present invention. The present invention may also be embodied as apparatus programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.

Claims

A calibration method for multiple groups of laser radar external parameters is characterized in that the multiple groups of laser radars comprise a first laser radar and a second laser radar, the first laser radar and the second laser radar are installed at different positions of a movable platform, and the calibration method comprises the following steps:

acquiring multiple frames of first point cloud data and multiple frames of second point cloud data which are respectively collected by the first laser radar and the second laser radar in a scanning calibration scene;

constructing a map according to the multiple frames of the first point cloud data and estimating a motion track of the movable platform relative to the calibration scene;

sequentially rotating and translating the multiple frames of second point cloud data into the map based on the motion tracks and the initial values of the external parameters so as to solve multiple groups of estimated external parameters and matching scores corresponding to each group of estimated external parameters;

determining a target outlier based on estimated outliers of at least a subset of the plurality of sets of estimated outliers, wherein a match score of each of the at least a subset of the sets of estimated outliers is greater than a threshold score.
The calibration method according to claim 1, wherein the plurality of frames of second point cloud data include second point cloud data of a current frame, the initial value of the extrinsic parameter includes a first initial value of the extrinsic parameter, and based on the motion trajectory and the initial value of the extrinsic parameter, the plurality of frames of second point cloud data are sequentially rotated and translated into the map to solve matching scores corresponding to the plurality of sets of estimated extrinsic parameters and each set of estimated extrinsic parameters, including:

based on the motion track and the first extrinsic parameter initial value, rotationally translating the second point cloud data of the current frame into the map to obtain a first converted point cloud of the current frame, wherein when the current frame is a first frame in the multiple frames, the first extrinsic parameter initial value is determined according to the position relation of the first laser radar and the second laser radar, or when the current frame is not the first frame, the first extrinsic parameter initial value is an estimated extrinsic parameter with a matching score higher than a threshold score, which is solved based on second point cloud data of a frame before the current frame;

acquiring a plurality of neighbor points in the map, wherein the distance between the neighbor points and the point cloud point in the first conversion point cloud is less than a threshold distance;

according to the first conversion point cloud and the plurality of neighbor points, a current group estimation extrinsic parameter and a matching score corresponding to the current frame are determined based on a geometric consistency constraint and a neighbor matching algorithm based on covariance weight matching, when the matching score of the current group estimation extrinsic parameter is higher than a threshold score, the current group estimation extrinsic parameter is used as the extrinsic parameter initial value for the next frame of the current frame, and when the matching score of the current group estimation extrinsic parameter is lower than the threshold score, the estimated extrinsic parameter with the matching score higher than the threshold score, which is solved based on second point cloud data of a frame before the current frame, is used as the extrinsic parameter initial value for the next frame.
The method of claim 2, wherein determining, from the first transformed point cloud and the plurality of neighbor points, a current set of estimated outliers and a matching score corresponding to the current frame based on a geometry consistency constraint and a neighbor matching algorithm based on covariance weight matching comprises:

determining an optimization equation based on geometric structure consistency constraint and a neighbor matching method based on covariance weight matching;

and substituting the first conversion point cloud and the plurality of neighbor points into the optimization equation to solve the current group of estimated external parameters and matching scores corresponding to the current frame.
The method of claim 3, wherein the optimization equation comprises a residual equation that is solved to obtain a current set of estimated outliers and the match score for the current frame, wherein the current set of estimated outliers minimizes an iterative residual of the residual equation, and wherein the smaller the iterative residual, the higher the match score.
The method of claim 3, wherein the weight is an eigenvalue of a covariance matrix of the plurality of neighbor points, wherein the plurality of neighbor points includes a first neighbor point and a second neighbor point, wherein the covariance of the first neighbor point is greater than the covariance of the second neighbor point, and wherein the weight of the first neighbor point is less than the weight of the second neighbor point.
The calibration method of claim 1, wherein determining a target external parameter based on the estimated external parameters of at least a portion of the plurality of sets of estimated external parameters comprises:

averaging the estimated external parameters of the at least partial group to obtain a target external parameter; or

Fitting the estimated external parameters of the at least partial group based on a least squares method to determine the target external parameters; or

Fitting the at least some groups of estimated external parameters based on a weighted least squares method with the matching scores as weights to determine the target external parameters.
A calibration method according to any one of claims 1 to 6, wherein the first lidar and the second lidar scan the calibration scene while the movable platform moves according to a predetermined motion trajectory in the calibration scene, or wherein the movable platform is fixed and the first lidar and the second lidar scan the calibration scene while the calibration scene is performed relative to the movable platform.
The calibration method according to any one of claims 1 to 7, wherein after determining that a coincidence rate of scanning trajectories of the first lidar and the second lidar in the calibration scene is greater than a preset threshold, a map is constructed according to the plurality of frames of the first point cloud data and a motion trajectory of the movable platform relative to the calibration scene is estimated.
A calibration method according to any one of claims 1 to 8, wherein the coincidence of the scanning trajectory of the first lidar and the scanning trajectory of the second lidar is above a threshold coincidence.
Calibration method according to any one of claims 1 to 9, wherein the external reference is a calibration matrix converting the coordinate system of the second lidar to the coordinate system of the first lidar, the calibration matrix comprising 3 rotation parameters and 3 translation parameters.
A calibration method according to any one of claims 1 to 10, further comprising:

rotationally registering the multi-frame second point cloud data to a coordinate system of the first laser radar based on the target external parameters;

and performing fusion operation on the rotationally registered multiple frames of second point cloud data and the multiple frames of first point cloud data to obtain point cloud data under the coordinate system of the first laser radar.
A calibration method according to any one of claims 1 to 11, wherein the field of view of the first lidar is less than 100 °; and/or the field of view of the second lidar is less than 100 °.
A calibration device for multiple groups of laser radar external parameters is characterized by comprising:

a movable platform;

the laser radar system comprises a movable platform body, a plurality of groups of laser radars, a plurality of laser radar sensors and a control module, wherein the plurality of groups of laser radars comprise a first laser radar and a second laser radar which are arranged at different positions of the movable platform body;

a memory having stored thereon a computer program for execution by the processor, and a processor for:

acquiring multiple frames of first point cloud data and multiple frames of second point cloud data which are respectively collected by the first laser radar and the second laser radar in a scanning calibration scene;

constructing a map according to the multiple frames of the first point cloud data and estimating a motion track of the movable platform relative to the calibration scene;

sequentially rotating and translating the multiple frames of second point cloud data into the map based on the motion tracks and the initial values of the external parameters so as to solve multiple groups of estimated external parameters and matching scores corresponding to each group of estimated external parameters;

determining a target outlier based on estimated outliers of at least a subset of the plurality of sets of estimated outliers, wherein a match score of each of the at least a subset of the sets of estimated outliers is greater than a threshold score.
Calibration arrangement according to claim 13, further comprising a calibration scenario configured to be movable outside the movable platform according to a predetermined motion trajectory.
The calibration apparatus according to claim 13, wherein the plurality of frames of second point cloud data includes second point cloud data of a current frame, the initial value of the extrinsic parameter includes a first initial value of the extrinsic parameter, and the processor is configured to sequentially rotate and translate the plurality of frames of second point cloud data into the map based on the motion trajectory and the initial value of the extrinsic parameter to solve matching scores corresponding to a plurality of sets of estimated extrinsic parameters and each set of estimated extrinsic parameters, including:

based on the motion track and the first extrinsic parameter initial value, rotationally translating the second point cloud data of the current frame into the map to obtain a first converted point cloud of the current frame, wherein when the current frame is a first frame in the multiple frames, the first extrinsic parameter initial value is determined according to the position relation of the first laser radar and the second laser radar, or when the current frame is not the first frame, the first extrinsic parameter initial value is an estimated extrinsic parameter with a matching score higher than a threshold score, which is solved based on second point cloud data of a frame before the current frame;

acquiring a plurality of neighbor points in the map, wherein the distance between the neighbor points and the point cloud point in the first conversion point cloud is less than a threshold distance;

according to the first conversion point cloud and the plurality of neighbor points, a current group estimation outlier and a matching score corresponding to the current frame are determined based on a geometric consistency constraint and a neighbor matching algorithm based on covariance weight matching, when the matching score of the current group estimation outlier is higher than a threshold score, the current group estimation outlier is used as the initial outlier value by the next frame of the current frame, and when the matching score of the current group estimation outlier is lower than the threshold score, the estimated outlier with the matching score higher than the threshold score, which is solved based on second point cloud data of a frame before the current frame, is used as the initial outlier value of the next frame by the next frame of the current frame.
The calibration apparatus of claim 15, wherein determining, from the first transformed point cloud and the plurality of neighbor points, a current set of estimated outliers and a matching score corresponding to the current frame based on a geometry consistency constraint and a neighbor matching algorithm based on covariance weight matching comprises:

determining an optimization equation based on geometric structure consistency constraint and a neighbor matching method based on covariance weight matching;

and substituting the first conversion point cloud and the plurality of neighbor points into the optimization equation to solve the current group of estimated external parameters and matching scores corresponding to the current frame.
The calibration apparatus as defined in claim 16, wherein the optimization equation includes a residual equation, and the current set of estimated parameters corresponding to the current frame and the matching score are obtained by solving the residual equation, wherein the current set of estimated parameters minimizes iterative residuals of the residual equation, and the iterative residuals are used to characterize the matching score.
The calibration apparatus as defined in claim 16, wherein the weight is an eigenvalue of a covariance matrix of the plurality of neighboring points, wherein the plurality of neighboring points includes a first neighboring point and a second neighboring point, and wherein the covariance of the first neighboring point is greater than the covariance of the second neighboring point, and the weight of the first neighboring point is less than the weight of the second neighboring point.
The calibration apparatus as defined in claim 13, wherein determining a target external parameter based on the estimated external parameters of at least a portion of the plurality of sets of estimated external parameters comprises:

averaging the estimated external parameters of the at least partial group to obtain a target external parameter; or

Fitting the estimated external parameters of the at least partial group based on a least squares method to determine the target external parameters; or

Fitting the at least some groups of estimated external parameters based on a weighted least squares method with the matching scores as weights to determine the target external parameters.
Calibration arrangement according to any one of claims 13 to 19, wherein the movable platform is adapted to scan the calibration scenario by the first lidar and the second lidar while moving according to a predetermined trajectory in the calibration scenario.
A calibration arrangement according to any one of claims 13 to 20, wherein the processor is configured to:

and after the coincidence rate of the scanning tracks of the first laser radar and the second laser radar in the calibration scene is determined to be greater than a preset threshold value, constructing a map according to the multiple frames of the first point cloud data and estimating the motion track of the movable platform relative to the calibration scene.
Calibration arrangement according to any one of claims 13 to 21, wherein the coincidence of the scanning trajectory of the first lidar and the scanning trajectory of the second lidar is above a threshold coincidence.
Calibration arrangement according to any one of claims 13 to 22, wherein the external reference is a calibration matrix converting the coordinate system of the second lidar to the coordinate system of the first lidar, the calibration matrix comprising 3 rotation parameters and 3 translation parameters.
The calibration apparatus as defined in claim 13, wherein the processor is further configured to:

rotationally registering the multi-frame second point cloud data to a coordinate system of the first laser radar based on the target external parameters;

and performing fusion operation on the rotationally registered multiple frames of second point cloud data and the multiple frames of first point cloud data to obtain point cloud data under the coordinate system of the first laser radar.
Calibration arrangement according to any one of claims 13-24, wherein the field of view of the first lidar is less than 100 °; and/or the field of view of the second lidar is less than 100 °.
A computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for calibrating multiple sets of lidar parameters of any of claims 1 to 12.