阅读说明：本技术 外参标定方法、装置、计算设备以及计算机存储介质 (External parameter calibration method and device, computing equipment and computer storage medium ) 是由 郭磊明 张莹莹 于 2019-08-21 设计创作，主要内容包括：本发明实施例公开了一种外参标定方法、装置、计算设备以及计算机存储介质,其中,外参标定方法包括：至少采集移动或旋转的第一激光雷达的第一点云序列以及第二激光雷达的第二点云序列,其中第一点云序列位于第一激光雷达的机体坐标系,第二点云序列位于第二激光雷达的机体坐标系；根据第一点云序列和第二点云序列分别获得预设时刻的第一激光雷达的第一点云地图以及第二激光雷达的第二点云地图；应用点云配准算法计算第二点云地图变换至第一点云地图的变换关系,获得标定结果。由此可见,利用本发明方案,使重叠区域不足的多激光雷达能够使用通用点云配准算法进行外参标定。(The embodiment of the invention discloses external reference calibration methods, devices, computing equipment and computer storage media, wherein the external reference calibration method comprises the steps of at least collecting point cloud sequences of a th laser radar and second point cloud sequences of a second laser radar which move or rotate, wherein the th point cloud sequence is located in a th laser radar body coordinate system, the second point cloud sequences are located in a second laser radar body coordinate system, a th point cloud map of the th laser radar and a second point cloud map of the second laser radar at preset time are respectively obtained according to the th point cloud sequences and the second point cloud sequences, a point cloud registration algorithm is applied to calculate a conversion relation from the second point cloud map to the th point cloud map, and a calibration result is obtained.)

1, method for calibrating external parameters, the method comprising:

collecting at least a point cloud sequence of a th laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the th laser radar and the second point cloud sequence is located in a body coordinate system of the second laser radar;

respectively obtaining a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset time according to the point cloud sequence and the second point cloud sequence;

and calculating a transformation relation from the second point cloud map to the th point cloud map by using a point cloud registration algorithm to obtain a calibration result.

2. The external reference calibration method according to claim 1, wherein the th lidar moved or rotated and the second lidar are located in a field having structured features.

3. The external reference calibration method according to claim 1, wherein the obtaining of the point cloud map of the laser radar at a preset time according to the point cloud sequence comprises:

applying a simultaneous localization and mapping algorithm to calculate a transformation relation of adjacent point clouds in the point cloud sequence;

transforming the th point cloud sequence to the th laser radar body coordinate system at the preset moment according to the transformation relation of adjacent point clouds to form a new th point cloud sequence;

merging the new point cloud sequences to obtain the point cloud map of the laser radar body coordinate system based on the preset time;

the obtaining of the second point cloud map of the second laser radar at a preset time according to the second point cloud sequence includes:

applying a simultaneous localization and mapping algorithm to calculate a transformation relationship of adjacent point clouds in the second point cloud sequence;

transforming the second point cloud sequence to the body coordinate system of the second laser radar at the preset moment according to the transformation relation of the adjacent point clouds to form a new second point cloud sequence;

and merging the new second point cloud sequences to obtain a second point cloud map of the body coordinate system of the second laser radar based on the preset moment, wherein the second point cloud map is partially overlapped with the th point cloud map.

4. The method for external reference calibration according to claim 3, wherein said applying a simultaneous localization and mapping algorithm to calculate the transformation relationship between adjacent point clouds in the -th point cloud sequence comprises:

traversing the point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the point cloud sequence by applying a point cloud registration algorithm;

optimizing the transformation relation of any adjacent point clouds in the point cloud sequence by using a general map optimization algorithm;

the applying a simultaneous localization and mapping algorithm to calculate a transformation relationship of adjacent point clouds in the second point cloud sequence comprises:

traversing the second point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the second point cloud sequence by applying a point cloud registration algorithm;

and optimizing the transformation relation of any adjacent point clouds in the second point cloud sequence by using a universal map optimization algorithm.

5. The extrinsic calibration method according to claim 3, wherein said new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new coordinates of the ith point cloud in the th point cloud sequence, i is a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iThe adjacent point clouds P of the point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)；

The new second point cloud sequence satisfies the following relation:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j)。

6. The external reference calibration method according to claim 1, wherein the applying the point cloud registration algorithm to calculate the transformation relationship from the second point cloud map to the -th point cloud map comprises, before obtaining the calibration result:

and respectively carrying out filtering operation on the point cloud map and the second point cloud map.

7. The extrinsic reference calibration method according to any of claims 1-6, wherein the point cloud registration algorithm comprises an iterative closest point algorithm or a normal distribution transformation algorithm.

8, external reference calibration device, characterized in that, external reference calibration device includes:

the data acquisition unit is used for at least acquiring a point cloud sequence of a th laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the th laser radar, and the second point cloud sequence is located in a body coordinate system of the second laser radar;

a registration unit, configured to apply a simultaneous localization and mapping algorithm, and obtain a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset times according to the point cloud sequence and the second point cloud sequence, respectively;

and the calibration unit is used for calculating the transformation relation from the second point cloud map to the th point cloud map by applying a point cloud registration algorithm to obtain a calibration result.

9, kinds of computing equipment, which is characterized in that the computing equipment comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface are communicated with each other through the communication bus;

the memory is configured to store at least executable instructions that cause the processor to perform the steps of the extrinsic calibration method according to any one of claims 1-7 through .

10, computer storage media having stored therein at least executable instructions for causing a processor to perform the steps of the extrinsic calibration method according to any one of claims 1-7 through .

Technical Field

The invention relates to the technical field of laser radars, in particular to external parameter calibration methods, devices, computing equipment and computer storage media.

Background

Lidar (light detection and ranging, lidar) is a optical remote sensing technology which measures parameters such as the distance of a target by irradiating beams of pulsed laser to the target, a set of three-dimensional measurement points obtained by the lidar may be called point cloud (point cloud) because the data volume is large and dense, point cloud registration is to find a transformation relation between two point set spaces given two three-dimensional data point sets from different coordinate systems so that the two point sets can be unified into a same coordinate system, and external reference calibration is used to determine the rotation and translation relation between multiple sensor coordinate systems in order to represent the multiple sensor data in the coordinate system of the system .

The point cloud registration algorithm is based on the premise that two point clouds have overlapping parts, for example, objects are irradiated by two laser radars together, in the unmanned application, the situation that the overlapping parts of the point clouds among a plurality of radars are few or do not have the overlapping parts at all exists, for example, laser radars are arranged in front of a vehicle head, radars are arranged at the tail of the vehicle, and due to the shielding of the vehicle body, the two radars do not have overlapping areas at all.

Disclosure of Invention

In view of the above, embodiments of the present invention provide methods, apparatuses, computing devices and computer storage media for external reference calibration that overcome or at least partially solve the above problems.

According to aspects of the invention, external reference calibration methods are provided, and the method comprises the steps of at least collecting a point cloud sequence of a movable or rotatable laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the laser radar, the second point cloud sequence is located in a body coordinate system of the second laser radar, obtaining a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset moments according to the point cloud sequence and the second point cloud sequence, and calculating a transformation relation of the second point cloud map transformed to the point cloud map by using a point cloud registration algorithm to obtain a calibration result.

Optionally, the th lidar moved or rotated and the second lidar moved or rotated are located in a field having structured features.

Optionally, the obtaining of the th lidar point cloud map at the preset time according to the th point cloud sequence includes applying a simultaneous localization and mapping algorithm to calculate a transformation relation between adjacent point clouds in the th point cloud sequence, transforming the th point cloud sequence to the th lidar body coordinate system at the preset time according to the transformation relation between the adjacent point clouds to form a new th point cloud sequence, merging the new th point cloud sequence to obtain the th lidar body coordinate system based on the th lidar body coordinate system at the preset time, obtaining the second lidar point cloud map at the preset time according to the second point cloud sequence, including applying the simultaneous localization and mapping algorithm to calculate the transformation relation between the adjacent point clouds in the second point cloud sequence, transforming the second point cloud sequence to the second lidar body coordinate system at the preset time according to the transformation relation between the adjacent point clouds to form a new second lidar body point cloud sequence, and overlapping the second lidar body coordinate system at the preset time with the second lidar body coordinate system at the preset time to obtain the merged map .

Optionally, the applying the simultaneous localization and mapping algorithm to calculate the transformation relationship between the adjacent point clouds in the th point cloud sequence includes traversing the th point cloud sequence, applying the point cloud registration algorithm to calculate the transformation relationship between any adjacent point clouds in the th point cloud sequence, using a general map optimization algorithm to optimize the transformation relationship between any adjacent point clouds in the th point cloud sequence, and applying the simultaneous localization and mapping algorithm to calculate the transformation relationship between any adjacent point clouds in the second point cloud sequence, including traversing the second point cloud sequence, applying the point cloud registration algorithm to calculate the transformation relationship between any adjacent point clouds in the second point cloud sequence, and using the general map optimization algorithm to optimize the transformation relationship between any adjacent point clouds in the second point cloud sequence.

Optionally, the new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new coordinates of the ith point cloud in the th point cloud sequence, i is a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)；

The new second point cloud sequence satisfies the following relation:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j)。

Optionally, the calculating, by using the point cloud registration algorithm, a transformation relationship from the second point cloud map to the th point cloud map includes performing filtering operations on the th point cloud map and the second point cloud map respectively before obtaining the calibration result.

Optionally, the point cloud registration algorithm includes an iterative closest point algorithm or a normal distribution transformation algorithm.

According to another aspects of the invention, external reference calibration devices are provided, which comprise a data acquisition unit, a registration unit and a calibration unit, wherein the data acquisition unit is used for at least acquiring a point cloud sequence of a moving or rotating 0 laser radar and a second point cloud sequence of a second laser radar, the point cloud sequence is located in a body coordinate system of the laser radar, the second point cloud sequence is located in a body coordinate system of the second laser radar, the registration unit is used for respectively acquiring an point cloud map of the laser radar and the second point cloud map of the second laser radar at a preset time according to the point cloud sequence and the second point cloud sequence, and the calibration unit is used for calculating a transformation relation of the second point cloud map to the point cloud map by applying a point cloud registration algorithm to obtain a calibration result.

According to another aspects of the invention, there is provided computing device comprising a processor, a memory, a communication interface, and a communication bus through which the processor, the memory, and the communication interface communicate with each other, the memory storing at least executable instructions for causing the processor to perform the steps of the aforementioned extrinsic parameter calibration method.

According to another aspects of the invention, computer storage media are provided having stored therein at least executable instructions that cause a processor to perform the steps of the aforementioned extrinsic calibration method.

In the embodiment of the invention, the external reference calibration method comprises the steps of at least collecting a point cloud sequence of a mobile or rotary th laser radar and a second point cloud sequence of a second laser radar, wherein the th point cloud sequence is located in a machine body coordinate system of the th laser radar, the second point cloud sequence is located in a machine body coordinate system of the second laser radar, a point cloud map of the th laser radar and a second point cloud map of the second laser radar at preset moments are respectively obtained according to the th point cloud sequence and the second point cloud sequence, a point cloud registration algorithm is applied to calculate the transformation relation of the second point cloud map to the th point cloud map, calibration results are obtained, therefore, the view field of a single laser radar is expanded through the movement or rotation of a carrier loaded with the laser radars, the point clouds of different coordinate systems at different moments are transformed to the same coordinate systems, the time registration of the multiple laser radars is carried out, the relative coordinate transformation relation between the maps is calculated through the respective machine body coordinates at the same moments, the external reference radar external point cloud point registration algorithm is achieved, and the external reference radar external reference calibration method can be used for calibrating the external reference radar under the common laser radar.

Drawings

the various embodiments are illustrated by way of example in the accompanying drawings and not by way of limitation, in which elements having the same reference number designation may be referred to by similar elements in the drawings and, unless otherwise indicated, the drawings are not to scale.

FIG. 1 is a flow chart illustrating an external reference calibration method according to an embodiment of the invention;

FIG. 2 is a schematic flow chart diagram illustrating another external reference calibration method according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a point cloud before calibration of external reference calibration methods according to an embodiment of the invention;

FIG. 4 is a schematic diagram illustrating two calibrated point clouds of laser radars according to external reference calibration methods in an embodiment of the invention;

FIG. 5 is a schematic diagram illustrating a point cloud after calibration of three laser radars according to external reference calibration methods in an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of another external reference calibration devices according to an embodiment of the present invention;

FIG. 7 illustrates a schematic structural diagram of computing devices according to an embodiment of the invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be 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 disclosure to those skilled in the art.

FIG. 1 is a schematic flow chart diagram illustrating external reference calibration methods according to an embodiment of the present invention, as shown in FIG. 1, the external reference calibration method includes:

step S11, at least a point cloud sequence of a laser radar and a second point cloud sequence of a second laser radar are collected, wherein the point cloud sequence is located in an laser radar body coordinate system, and the second point cloud sequence is located in the second laser radar body coordinate system.

In step S11, the laser radar and the second laser radar which move or rotate are located in a field with a structural feature, in the embodiment of the present invention, a field including abundant structural features is selected, a carrier carrying at least the laser radar and the second laser radar is moved or rotated on the field, the point cloud sequence of the 0 laser radar and the second point cloud sequence of the second laser radar are simultaneously collected, in the case of two laser radars, fields which are rich in wall and shaft are selected, the carrier carrying the laser radar and the second laser radar advances by 10 meters at a constant speed and straight line at a speed of 5km/h, and then travels according to the 8298 word while collecting the point cloud sequences of the two laser radars, in the body coordinate system of the laser radar, preferably, the body coordinate system of the laser radar may be a coordinate system with the laser center as the origin, and the point cloud sequence of the L is represented as L₁＝{P₀，P₁,…P_nN is a positive integer. The second point cloud sequence is located in the body coordinate system of the second laser radar, preferably, the body coordinate system of the second laser radar may be a coordinate system with a second laser radar center as an origin, and the second point cloud sequence is represented by L2 ═ P0, P1, … P_mAnd m is a positive integer.

In the embodiment of the invention, a plurality of laser radars can also be rigidly arranged on an carrier, the carrier runs at low speed and uniform speed, the motion track comprises a straight line or a curve, and a point cloud sequence of each laser radar is collected.

And S12, respectively obtaining a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset time according to the point cloud sequence and the second point cloud sequence.

The preset time may be the initial time of the th point cloud sequence and the second point cloud sequence, or may be any other time of the selected th point cloud sequence and the second point cloud sequence in the acquisition process, which is not limited herein.

The th point cloud sequence and the second point cloud sequence are respectively explained below, and in the embodiment of the present invention, as shown in fig. 2, for the th point cloud sequence, the step S12 includes:

and step S121, applying a simultaneous localization and mapping algorithm to calculate the transformation relation of the adjacent point clouds in the point cloud sequence.

Specifically, traversing the point cloud sequence, calculating the transformation relation of any adjacent point clouds in the point cloud sequence by applying a point cloud registration algorithm, optimizing the transformation relation of any adjacent point clouds in the point cloud sequence by using a universal map optimization algorithm, wherein the transformation relation of the adjacent point clouds can be the simple coordinate transformation relation of the adjacent point clouds and can also be other transformation relations, such as scaling and the like, and for the point cloud sequence L₁From 1 to n, a Point cloud Simultaneous localization and mapping (SLAM) algorithm is used, for example, a Normal Distribution Transformation (NDT) algorithm or an Iterative Closest Point (ICP) algorithm is used first, then a General Graph Optimization (G2O) algorithm is used for Optimization, and a th Point cloud sequence L is obtained by calculation₁Coordinate transformation relation T of middle adjacent point clouds_i＝F(P_i-1，P_i) Wherein P is_i-1Is the coordinates of the (i-1) th point cloud in the th point cloud sequence, P_iIs the coordinates of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iThe transformation relationship of (1).

And S122, transforming the th point cloud sequence to the th laser radar body coordinate system at the preset moment according to the coordinate transformation relation of adjacent point clouds to form a new th point cloud sequence.

In the embodiment of the invention, the new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new coordinates of the ith point cloud in the th point cloud sequence, i is a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)。

As shown in FIG. 3, a new point cloud sequence before calibration is obtained and is denoted as L'₁＝{P'₀，P'₁,…P'_nDenoted L 'as a new second point cloud sequence'₂＝{P'₀，P'₁,…P'_m}。

In the embodiment of the invention, preferably, the starting time is selected, and the th point cloud sequence and the second point cloud sequence are transformed to the body coordinate system of the corresponding laser radar at the starting time, namely, the th point cloud sequence is transformed to the body coordinate system of the th laser radar at the starting time, and the second point cloud sequence is transformed to the body coordinate system of the second laser radar at the starting time.

And S123, merging the new th point cloud sequences to obtain the th point cloud map of the th laser radar body coordinate system based on the preset time.

The method comprises the steps of calculating a transformation relation of any adjacent point clouds in a second point cloud sequence by traversing the second point cloud sequence and applying a point cloud registration algorithm, optimizing the transformation relation of any adjacent point clouds in the second point cloud sequence by using a universal map optimization algorithm, and performing L-mapping on the second point cloud sequence₂From 1 to m, point cloud sequence is calculated by using SLAM algorithmColumn L₂Coordinate transformation relation T of middle adjacent point clouds_jF (Pj-1, Pj), wherein P_j-1Is the coordinates of the j-1 point cloud in the th point cloud sequence, P_jIs the coordinate of the jth point cloud in the th point cloud sequence, T_jFor the adjacent point cloud P in the th point cloud sequence_j-1And P_jThe transformation relationship of (1).

And then combining the new second point cloud sequences to obtain a second point cloud map based on the body coordinate system of the second laser radar at the preset time, wherein the second point cloud map is partially overlapped with the point cloud map, wherein the new second point cloud sequence satisfies the following relational expression:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j). Obtaining a new second point cloud sequence before calibration as L'₂＝{P'₀，P'₁,…P'_m}. New point cloud sequence L 'before calibration'₁And a new second point cloud sequence L 'before calibration'₂A single frame point cloud of (a) is shown in fig. 3.

Combining n point clouds in a new point cloud sequence before calibration to obtain a th point cloud map M of a th laser radar body coordinate system based on the previously selected preset time₁ th point cloud map M₁Corresponding to the superposition of n point clouds in the point cloud sequence, and the shape of the point cloud sequence is L 'in figure 3'₁Similarly, but the dots are more dense. Correspondingly, combining the M point clouds in the new second point cloud sequence to obtain a second point cloud map M of the body coordinate system of the second laser radar before calibration based on the previously selected preset time₂. Second point cloud map M₂Corresponding to the superposition of m point clouds in the second point cloud sequence, the shape of which is L 'in figure 3'₂Similar, but more densely populated Point cloud map M₁And a second point cloud map M₂Partially overlapping.

And S13, calculating the transformation relation from the second point cloud map to the point cloud map by using a point cloud registration algorithm to obtain a calibration result.

Specifically, a point cloud registration algorithm is applied to calculate a second point cloud map M₂Transformation to th point cloud map M₁I.e. the transformation matrix T ═ F (M)₁,M₂) The point cloud registration algorithm comprises an iterative closest point algorithm (ICP) or a normal distribution transformation algorithm (NDT), and in other embodiments of the invention, other point cloud registration algorithms can be applied, without limitation₁Transform to the second point cloud map M₂The transformation relationship of (1). According to the calibration result, the second point cloud map M₂And th point cloud map M₁Calibrating to obtain a calibrated point cloud map, wherein the calibration process is equivalent to that of the th point cloud map M₁And a second point cloud map M₂Transforming to the same coordinate system, and processing the map M by matching the point cloud₁And a second point cloud map M₂The calibrated point cloud map obtained after calibration is equivalent to the th point cloud map M₁Or a second point cloud map M₂The distance offset and/or the angle of rotation relative to another pieces of material are such that the two overlap as much as possible.

In order to accelerate the calculation, before step S13, filtering operations such as thinning, denoising, feature point extraction, and the like are respectively performed on the obtained point cloud map and the obtained second point cloud map, specifically, filtering operations such as thinning, denoising, feature point extraction, and the like are respectively performed on the point cloud map and the second point cloud map, so as to reduce the data volume, improve the data quality, and facilitate subsequent calibration.

In the embodiment of the invention, a second point cloud map M is obtained₂Relative th point cloud map M₁After the calibration matrix T is calibrated, the calibration matrix T can be directly applied to the second point cloud sequence L of the second laser radar₂And point cloud sequence L of th laser radar₁Calibration is performed, and FIG. 4 is a second point cloud sequence L of the second laser radar₂And point cloud sequence L of th laser radar₁And (5) calibrating the point cloud.

It should be noted that, when there are a plurality of laser radars, calibration matrices of the point cloud sequence of each laser radar with respect to the point cloud sequences of laser radars may be obtained, and then each calibration matrix is calibrated₁、L₂、L₃Firstly, respectively obtaining point cloud sequences L of the second laser radar₂Point cloud sequence L for th lidar₁And the point cloud sequence L of the third laser radar₃Point cloud sequence L for th lidar₁And then calibrating the point cloud sequences of the three laser radars to obtain the calibrated point cloud. Or the calibration matrix of the point cloud sequence of any two laser radars can be obtained first, and then each calibration matrix is calibrated. E.g. point cloud sequence L for three laser radars₁、L₂、L₃Firstly, respectively obtaining point cloud sequences L of the second laser radar₂Point cloud sequence L for th lidar₁And the point cloud sequence L of the third laser radar₃Point cloud sequence L relative to the second lidar₂And then calibrating the point cloud sequences of the three laser radars according to the two calibration matrixes to obtain calibrated point clouds. The calibrated point clouds finally obtained by the two methods are the same. FIG. 5 is a point cloud sequence L for three lidar₁、L₂、L₃And (5) calibrating the point cloud.

In the embodiment of the invention, the external reference calibration method comprises the steps of at least collecting a point cloud sequence of a mobile or rotary th laser radar and a second point cloud sequence of a second laser radar, wherein the th point cloud sequence is located in a body coordinate system of the th laser radar, the second point cloud sequence is located in a body coordinate system of the second laser radar, a point cloud map of the th laser radar and a second point cloud map of the second laser radar at preset moments are respectively obtained according to the th point cloud sequence and the second point cloud sequence, a point cloud registration algorithm is applied to calculate the transformation relation of the second point cloud map to the th point cloud map, calibration results are obtained, therefore, the view field of a single laser radar is expanded through the movement or rotation of a carrier loaded with the laser radars, the point clouds of different coordinate systems at different moments are transformed to the same coordinate systems, the time registration of the multiple laser radars is carried out, the relative coordinate transformation relation between the maps is calculated through the respective body coordinates at the same moments, the external reference radar external point cloud registration algorithm is realized, and the external reference radar can be calibrated through the laser radar under the multiple laser radar.

Fig. 6 shows a schematic structural diagram of an external reference calibration apparatus according to an embodiment of the present invention. As shown in fig. 6, the external reference calibration apparatus includes: a data acquisition unit 601, a registration unit 602, and a calibration unit 603. Wherein:

the data acquisition unit 601 is used for at least acquiring a point cloud sequence of a th moving or rotating laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the th laser radar, the second point cloud sequence is located in a body coordinate system of the second laser radar, the registration unit 602 is used for respectively obtaining a point cloud map of the th laser radar and a second point cloud map of the second laser radar at a preset moment according to the th point cloud sequence and the second point cloud sequence, and the calibration unit 603 is used for calculating a transformation relation of the second point cloud map to the th point cloud map by applying a point cloud registration algorithm to obtain a calibration result.

In alternative, the th lidar that moves or rotates and the second lidar are located in a field having structured features.

In optional manners, the registration unit 602 is configured to apply a simultaneous localization and mapping algorithm to respectively calculate a transformation relationship between adjacent point clouds in the point cloud sequence, select a preset time, respectively transform the point cloud sequence to the laser radar body coordinate system at the preset time according to the coordinate transformation relationship between the adjacent point clouds to form a new point cloud sequence, merge the new point cloud sequence to obtain the point cloud map based on the laser radar body coordinate system at the preset time, apply a simultaneous localization and mapping algorithm to calculate a transformation relationship between the adjacent point clouds in the second point cloud sequence, transform the second point cloud sequence to the second laser radar body coordinate system at the preset time according to the transformation relationship between the adjacent point clouds to form a new second point cloud sequence, and merge the new second point cloud sequence to obtain the second laser radar body map based on the second laser radar at the preset time, where the second point cloud map and the second point cloud map are partially overlapped .

In optional manners, the registration unit 602 is further configured to traverse the point cloud sequence, calculate a transformation relationship between any neighboring point clouds in the point cloud sequence by using a point cloud registration algorithm, optimize a transformation relationship between any neighboring point clouds in the point cloud sequence by using a general purpose map optimization algorithm, traverse the second point cloud sequence, calculate a transformation relationship between any neighboring point clouds in the second point cloud sequence by using a point cloud registration algorithm, and optimize a transformation relationship between any neighboring point clouds in the second point cloud sequence by using a general purpose map optimization algorithm.

In alternative modes, the new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new point cloudCoordinates of the ith point cloud in the sequence, i being a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)；

The new second point cloud sequence satisfies the following relation:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j)。

In , the registration unit 602 is further configured to perform a filtering operation on the point cloud map and the second point cloud map, respectively.

In alternative, the point cloud registration algorithm includes an iterative closest point algorithm or a normal distribution transformation algorithm.

In the embodiment of the invention, the external reference calibration method comprises the steps of at least collecting a point cloud sequence of a mobile or rotary th laser radar and a second point cloud sequence of a second laser radar, wherein the th point cloud sequence is located in a machine body coordinate system of the th laser radar, the second point cloud sequence is located in a machine body coordinate system of the second laser radar, a point cloud map of the th laser radar and a second point cloud map of the second laser radar at preset moments are respectively obtained according to the th point cloud sequence and the second point cloud sequence, a point cloud registration algorithm is applied to calculate the transformation relation of the second point cloud map to the th point cloud map, calibration results are obtained, therefore, the view field of a single laser radar is expanded through the movement or rotation of a carrier loaded with the laser radars, the point clouds of different coordinate systems at different moments are transformed to the same coordinate systems, the time registration of the multiple laser radars is carried out, the relative coordinate transformation relation between the maps is calculated through the respective machine body coordinates at the same moments, the external reference radar external point cloud point registration algorithm is achieved, and the external reference radar external reference calibration method can be used for calibrating the external reference radar under the common laser radar.

The embodiment of the invention provides nonvolatile computer storage media, wherein the computer storage media store at least executable instructions, and the computer executable instructions can execute the external reference calibration method in any method embodiment.

The executable instructions may be specifically configured to cause the processor to:

collecting at least a point cloud sequence of a th laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the th laser radar and the second point cloud sequence is located in a body coordinate system of the second laser radar;

respectively obtaining a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset time according to the point cloud sequence and the second point cloud sequence;

and calculating a transformation relation from the second point cloud map to the th point cloud map by using a point cloud registration algorithm to obtain a calibration result.

In alternative, the th lidar that moves or rotates and the second lidar are located in a field having structured features.

In alternative forms, the executable instructions may be specifically configured to cause the processor to:

applying a simultaneous localization and mapping algorithm to calculate a transformation relation of adjacent point clouds in the point cloud sequence;

transforming the th point cloud sequence to the th laser radar body coordinate system at the preset moment according to the coordinate transformation relation of adjacent point clouds to form a new th point cloud sequence;

merging the new point cloud sequences to obtain the point cloud map of the laser radar body coordinate system based on the preset time;

applying a simultaneous localization and mapping algorithm to calculate a transformation relationship of adjacent point clouds in the second point cloud sequence;

transforming the second point cloud sequence to the body coordinate system of the second laser radar at the preset moment according to the transformation relation of the adjacent point clouds to form a new second point cloud sequence;

and merging the new second point cloud sequences to obtain a second point cloud map of the body coordinate system of the second laser radar based on the preset moment, wherein the second point cloud map is partially overlapped with the th point cloud map.

In alternative forms, the executable instructions may be specifically configured to cause the processor to:

traversing the point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the point cloud sequence by applying a point cloud registration algorithm;

optimizing a transformation relationship of any adjacent point clouds in the point cloud sequence and the second point cloud sequence by using a general map optimization algorithm;

traversing the second point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the second point cloud sequence by applying a point cloud registration algorithm;

and optimizing the transformation relation of any adjacent point clouds in the second point cloud sequence by using a universal map optimization algorithm.

In alternative modes, the new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new coordinates of the ith point cloud in the th point cloud sequence, i is a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)；

The new second point cloud sequence satisfies the following relation:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j)。

In alternative forms, the executable instructions may be specifically configured to cause the processor to:

and respectively carrying out filtering operation on the point cloud map and the second point cloud map.

In alternative, the point cloud registration algorithm includes an iterative closest point algorithm or a normal distribution transformation algorithm.

In the embodiment of the invention, the external reference calibration method comprises the steps of at least collecting a point cloud sequence of a mobile or rotary th laser radar and a second point cloud sequence of a second laser radar, wherein the th point cloud sequence is located in a machine body coordinate system of the th laser radar, the second point cloud sequence is located in a machine body coordinate system of the second laser radar, a point cloud map of the th laser radar and a second point cloud map of the second laser radar at preset moments are respectively obtained according to the th point cloud sequence and the second point cloud sequence, a point cloud registration algorithm is applied to calculate the transformation relation of the second point cloud map to the th point cloud map, calibration results are obtained, therefore, the view field of a single laser radar is expanded through the movement or rotation of a carrier loaded with the laser radars, the point clouds of different coordinate systems at different moments are transformed to the same coordinate systems, the time registration of the multiple laser radars is carried out, the relative coordinate transformation relation between the maps is calculated through the respective machine body coordinates at the same moments, the external reference radar external point cloud point registration algorithm is achieved, and the external reference radar external reference calibration method can be used for calibrating the external reference radar under the common laser radar.

An embodiment of the present invention provides computer program products comprising a computer program stored on a computer storage medium, the computer program comprising program instructions that, when executed by a computer, cause the computer to perform the method of extrinsic calibration in any of the above method embodiments.

The executable instructions may be specifically configured to cause the processor to:

collecting at least a point cloud sequence of a th laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the th laser radar and the second point cloud sequence is located in a body coordinate system of the second laser radar;

respectively obtaining a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset time according to the point cloud sequence and the second point cloud sequence;

and calculating a transformation relation from the second point cloud map to the th point cloud map by using a point cloud registration algorithm to obtain a calibration result.

In alternative, the th lidar that moves or rotates and the second lidar are located in a field having structured features.

In alternative forms, the executable instructions may be specifically configured to cause the processor to:

applying a simultaneous localization and mapping algorithm to calculate a transformation relation of adjacent point clouds in the point cloud sequence;

transforming the th point cloud sequence to the th laser radar body coordinate system at the preset moment according to the coordinate transformation relation of adjacent point clouds to form a new th point cloud sequence;

merging the new point cloud sequences to obtain the point cloud map of the laser radar body coordinate system based on the preset time;

applying a simultaneous localization and mapping algorithm to calculate a transformation relationship of adjacent point clouds in the second point cloud sequence;

transforming the second point cloud sequence to the body coordinate system of the second laser radar at the preset moment according to the transformation relation of the adjacent point clouds to form a new second point cloud sequence;

and merging the new second point cloud sequences to obtain a second point cloud map of the body coordinate system of the second laser radar based on the preset moment, wherein the second point cloud map is partially overlapped with the th point cloud map.

In alternative forms, the executable instructions may be specifically configured to cause the processor to:

traversing the point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the point cloud sequence by applying a point cloud registration algorithm;

optimizing the transformation relation of any adjacent point clouds in the point cloud sequence by using a general map optimization algorithm;

traversing the second point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the second point cloud sequence by applying a point cloud registration algorithm;

and optimizing the transformation relation of any adjacent point clouds in the second point cloud sequence by using a universal map optimization algorithm.

In alternative modes, the new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new coordinates of the ith point cloud in the th point cloud sequence, i is a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)；

The new second point cloud sequence satisfies the following relation:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j)。

In alternative forms, the executable instructions may be specifically configured to cause the processor to:

and respectively carrying out filtering operation on the point cloud map and the second point cloud map.

In alternative, the point cloud registration algorithm includes an iterative closest point algorithm or a normal distribution transformation algorithm.

In the embodiment of the invention, the external reference calibration method comprises the steps of at least collecting a point cloud sequence of a mobile or rotary th laser radar and a second point cloud sequence of a second laser radar, wherein the th point cloud sequence is located in a machine body coordinate system of the th laser radar, the second point cloud sequence is located in a machine body coordinate system of the second laser radar, a point cloud map of the th laser radar and a second point cloud map of the second laser radar at preset moments are respectively obtained according to the th point cloud sequence and the second point cloud sequence, a point cloud registration algorithm is applied to calculate the transformation relation of the second point cloud map to the th point cloud map, calibration results are obtained, therefore, the view field of a single laser radar is expanded through the movement or rotation of a carrier loaded with the laser radars, the point clouds of different coordinate systems at different moments are transformed to the same coordinate systems, the time registration of the multiple laser radars is carried out, the relative coordinate transformation relation between the maps is calculated through the respective machine body coordinates at the same moments, the external reference radar external point cloud point registration algorithm is achieved, and the external reference radar external reference calibration method can be used for calibrating the external reference radar under the common laser radar.

Fig. 7 is a schematic structural diagram of an embodiment of the apparatus according to the present invention, and the specific embodiment of the present invention does not limit the specific implementation of the apparatus.

As shown in fig. 7, the apparatus may include: a processor (processor)702, a communications interface 704, a memory 706, and a communications bus 708.

Wherein: the processor 702, communication interface 704, and memory 706 communicate with each other via a communication bus 708. A communication interface 704 for communicating with network elements of other devices, such as clients or other servers. The processor 702 is configured to execute the program 710, and may specifically execute the relevant steps in the above-described external reference calibration method embodiment.

In particular, the program 710 may include program code that includes computer operating instructions.

The processor 702 may be a central processing unit CPU, or an Application Specific Integrated Circuit (ASIC), or or more Integrated circuits configured to implement embodiments of the present invention, the device includes or more processors, which may be of the same type, such as or more CPUs, or different types of processors, such as or more CPUs and or more ASICs.

Memory 706 is used to store the program 710. memory 706 may comprise high-speed RAM memory, and may also include non-volatile memory, such as at least disk drives.

The program 710 may specifically be used to cause the processor 702 to perform the following operations:

collecting at least a point cloud sequence of a th laser radar and a second point cloud sequence of a second laser radar, wherein the point cloud sequence is located in a body coordinate system of the th laser radar and the second point cloud sequence is located in a body coordinate system of the second laser radar;

respectively obtaining a point cloud map of the laser radar and a second point cloud map of the second laser radar at preset time according to the point cloud sequence and the second point cloud sequence;

and calculating a transformation relation from the second point cloud map to the th point cloud map by using a point cloud registration algorithm to obtain a calibration result.

In alternative, the th lidar that moves or rotates and the second lidar are located in a field having structured features.

In alternative, the program 710 may be specifically configured to cause the processor 702 to:

applying a simultaneous localization and mapping algorithm to calculate a transformation relation of adjacent point clouds in the point cloud sequence;

transforming the th point cloud sequence to the th laser radar body coordinate system at the preset moment according to the coordinate transformation relation of adjacent point clouds to form a new th point cloud sequence;

merging the new point cloud sequences to obtain the point cloud map of the laser radar body coordinate system based on the preset time;

applying a simultaneous localization and mapping algorithm to calculate a transformation relationship of adjacent point clouds in the second point cloud sequence;

transforming the second point cloud sequence to the body coordinate system of the second laser radar at the preset moment according to the transformation relation of the adjacent point clouds to form a new second point cloud sequence;

and merging the new second point cloud sequences to obtain a second point cloud map of the body coordinate system of the second laser radar based on the preset moment, wherein the second point cloud map is partially overlapped with the th point cloud map.

In alternative, the program 710 may be specifically configured to cause the processor 702 to:

traversing the point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the point cloud sequence by applying a point cloud registration algorithm;

optimizing the transformation relation of any adjacent point clouds in the point cloud sequence by using a general map optimization algorithm;

traversing the second point cloud sequence, and calculating a transformation relation of any adjacent point clouds in the second point cloud sequence by applying a point cloud registration algorithm;

and optimizing the transformation relation of any adjacent point clouds in the second point cloud sequence by using a universal map optimization algorithm.

In alternative modes, the new point cloud sequence satisfies the following relation:

P′_i＝T₁T₂…T_iP_i，

wherein, P'_iIs the new coordinates of the ith point cloud in the th point cloud sequence, i is a positive integer, P_iIs the coordinate of the ith point cloud in the th point cloud sequence, T_iFor the adjacent point cloud P in the th point cloud sequence_i-1And P_iTransformation relation of (1), T_i＝F(P_i-1,P_i)；

The new second point cloud sequence satisfies the following relation:

P′_j＝T₁T₂…T_jP_j，

wherein, P'_jIs the new coordinate of the jth point cloud in the second point cloud sequence, j is a positive integer, P_jIs the coordinate, T, of the jth point cloud in the second point cloud sequence_jFor adjacent point clouds P in the second point cloud sequence_j-1And P_jTransformation relation of (1), T_j＝F(P_j-1,P_j)。

In alternative, the program 710 may be specifically configured to cause the processor 702 to:

and respectively carrying out filtering operation on the point cloud map and the second point cloud map.

In alternative, the point cloud registration algorithm includes an iterative closest point algorithm or a normal distribution transformation algorithm.

In the embodiment of the invention, the external reference calibration method comprises the steps of at least collecting a point cloud sequence of a mobile or rotary th laser radar and a second point cloud sequence of a second laser radar, wherein the th point cloud sequence is located in a th laser radar body coordinate system, the second point cloud sequence is located in a second laser radar body coordinate system, a point cloud map of a th laser radar and a second point cloud map of the second laser radar at preset moments are respectively obtained according to the th point cloud sequence and the second point cloud sequence, a point cloud registration algorithm is applied to calculate the transformation relation of the second point cloud map to the th point cloud map, calibration results are obtained, therefore, the view field of a single laser radar is expanded through the movement or rotation of a carrier loaded with the laser radar, the point clouds of different coordinate systems at different moments are transformed to the same coordinate systems, the time registration of the multiple laser radars is carried out, the relative coordinate transformation relation of the multiple laser radars is calculated by using respective body coordinates at the same moment, the point cloud point maps at different moments are subjected to realize the external reference radar external reference calibration, and the external reference radar is calibrated by using a multiple laser radar external reference radar.

The algorithms or displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus.

However, it is understood that embodiments of the invention may be practiced without these specific details, and that examples well-known methods, structures, and techniques have not been shown in detail in order not to obscure the understanding of this description.

Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together by in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of the various inventive aspects, however, the disclosed method is not intended to be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim.

It will be understood by those skilled in the art that modules in the apparatus of the embodiments may be adaptively changed and arranged in or more apparatuses different from the embodiments, that modules or units or components in the embodiments may be combined into modules or units or components, and further, that they may be divided into sub-modules or sub-units or sub-components, that all features disclosed in this specification (including the accompanying claims, abstract and drawings), and all processes or units of any method or apparatus so disclosed, may be combined in any combination, except at least of such features and/or processes or units are mutually exclusive, unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose.

Furthermore, those of skill in the art will appreciate that while the embodiments described herein include some features included in other embodiments, not others, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments.

The invention may be embodied by means of hardware comprising several distinct elements, and by means of a suitably programmed computer, in a unit claim enumerating several means, several of these means may be embodied by one and the same item of hardware, the use of the words , second, third, etc. may indicate any sequence.