阅读说明：本技术 一种基于rviz实现多激光雷达点云对齐的高效静态标定方法 (Efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz ) 是由 张林灿 董钊志 宋罡 张婉蒙 刘树全 于 2020-05-07 设计创作，主要内容包括：本发明公开了一种基于rviz实现多激光雷达点云对齐的高效静态标定方法,步骤如下：量测出各激光雷达相对车身坐标系的相对位置；将三个标定长方体,放置在车身坐标系中的坐标下,形成三角之势；然后编写一个话题发布程序,按照三个物体实际形状和放置位置画出它们在车身坐标系中的位置,启动话题发布程序和rviz,rviz监听多路激光雷达的点云和物体的话题,在三维视图中观察点云成像和所画的物体位置的差异；编写一个参数标定配置文件,包含六个参数,分别表示平移向量的三个参数和滚转角、俯仰角以及偏航角；通过修改三个角度值,观察点云成像移动趋势,直到调节到与rviz中画出的三个参照物重合即可；最后再微调一下平移向量的数值即可,整个标定过程完成。(The invention discloses an efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz, which comprises the following steps: measuring the relative position of each laser radar relative to the vehicle body coordinate system; placing the three calibration cuboids under the coordinates in the vehicle body coordinate system to form a triangular potential; then writing a topic publishing program, drawing the positions of the three objects in a vehicle body coordinate system according to the actual shapes and the placing positions of the three objects, starting the topic publishing program and rviz, monitoring the point cloud of a multi-path laser radar and the topic of the object by the rviz, and observing the difference between the point cloud imaging and the drawn object position in a three-dimensional view; compiling a parameter calibration configuration file, which comprises six parameters respectively representing three parameters of a translation vector, a rolling angle, a pitch angle and a yaw angle; observing the point cloud imaging movement trend by modifying the three angle values until the point cloud imaging movement trend is adjusted to be coincident with three reference objects drawn in rviz; and finally, fine-tuning the numerical value of the translation vector, and finishing the whole calibration process.)

1. An efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz is characterized by comprising the following steps: the method comprises the following steps:

the first step is as follows: measuring the relative position of each laser radar relative to the coordinate system of the vehicle body, namely the translation vector [ x ]_a,y_a,z_a]^T；

The second step is that: three calibration cuboids with the length of 20cm, the width of 20cm and the height of 200cm are respectively placed under coordinates (20m,5m,0m), (30m,0m,0m), (20m, -5m,0m) in a vehicle body coordinate system to form a triangular potential;

the third step: then writing a topic publishing program, drawing the positions of three objects in a vehicle body coordinate system according to the actual shapes and the placing positions of the three objects, starting the topic publishing program and rviz, monitoring the point cloud of a multi-path laser radar and the topic of the objects by the rviz, observing the difference between the point cloud imaging and the positions of the drawn objects in a three-dimensional view, writing a parameter calibration configuration file, wherein the parameter calibration configuration file comprises six parameters which are x, y, z, roll, pitch and yaw respectively, and the six parameters respectively represent three parameters of a translation vector, a roll angle, a pitch angle and a yaw angle;

the fourth step: observing the point cloud imaging movement trend by modifying the angle values of a rolling angle, a pitch angle and a yaw angle until the point cloud imaging movement trend is adjusted to be coincident with three reference objects drawn in rviz;

the fifth step: and finally, fine-tuning the numerical value of the translation vector, namely, the point cloud imaging of the three objects in the fourth step is overlapped with the three reference objects drawn in rviz, and the point clouds formed by any other objects in the common coverage area of the three laser radars are completely overlapped and consistent when moving or standing, so that the whole calibration process is completed.

The technical field is as follows:

the invention relates to an efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz, and belongs to the technical field of intelligent driving of electric vehicles.

Background art:

the Multi-sensor Information Fusion (MSIF) is an Information processing process which is carried out by utilizing computer technology to automatically analyze and synthesize Information and data from multiple sensors or multiple sources under a certain criterion so as to complete needed decision and estimation. Before the step of the above-mentioned operation is carried out, for multiple laser radars, in the installation process of each laser radar, due to the problem of the installation precision of a mechanical structure, the phenomena of angular deviation such as pitching, yawing, rolling and the like are inevitable, in addition, the installation positions of different laser radars are different, and all the factors determine that the coordinate system of each laser radar needs to be synchronized to the coordinate system of the vehicle body so as to meet the condition that the point clouds of different laser radars are uniquely determined for the identification of the same object; if the step is not completed, the same obstacle can be identified into a plurality of objects by the multi-path laser radar, and further misleading judgment of subsequent fusion decisions is caused, so that immeasurable results are caused. At present, the problem that a plurality of laser radars, especially products of different brands, are difficult to calibrate is particularly obvious in the use process, technicians of a plurality of manufacturers are often required to perform combined calibration, and most of laser radar manufacturers provide calibration at present, so that the calibration is paid, and the whole calibration is time-consuming, labor-consuming and expensive.

Therefore, there is a need to improve the prior art to overcome the deficiencies of the prior art.

The invention content is as follows:

the invention provides an efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz for solving the problems in the prior art, which can be used for synchronizing the coordinate systems of all radars to a vehicle body coordinate system when a plurality of laser radars are used as sensors of an intelligent driving vehicle, so that the alignment operation of the laser radar point cloud on original data is realized.

The technical scheme adopted by the invention is as follows: an efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz comprises the following steps:

the first step is as follows: measuring the relative position of each laser radar relative to the coordinate system of the vehicle body, namely the translation vector [ x ]_a,y_a,z_a]^T；

The second step is that: placing three calibration cuboids with the length of 20cm, the width of 20cm and the height of 200cm under coordinates (20m,5m,0m), (30m,0m,0m) and (20m, -5m,0m) in a vehicle body coordinate system respectively, wherein the coordinates correspond to (x, y, z) in the vehicle body coordinate system respectively to form a triangular potential;

the third step: then writing a topic publishing program, drawing the positions of three objects in a vehicle body coordinate system according to the actual shapes and the placing positions of the three objects, starting the topic publishing program and rviz, monitoring the point cloud of a multi-path laser radar and the topic of the objects by the rviz, observing the difference between the point cloud imaging and the positions of the drawn objects in a three-dimensional view, writing a parameter calibration configuration file, wherein the parameter calibration configuration file comprises six parameters which are x, y, z, roll, pitch and yaw respectively, and the six parameters respectively represent three parameters of a translation vector, a roll angle, a pitch angle and a yaw angle;

the fourth step: observing the point cloud imaging movement trend by modifying the angle values of a rolling angle, a pitch angle and a yaw angle until the point cloud imaging movement trend is adjusted to be coincident with three reference objects drawn in rviz;

the fifth step: and finally, fine-tuning the numerical value of the translation vector, namely, the point cloud imaging of the three objects in the fourth step is overlapped with the three reference objects drawn in rviz, and the point clouds formed by any other objects in the common coverage area of the three laser radars are completely overlapped and consistent when moving or standing, so that the whole calibration process is completed.

The invention has the following beneficial effects: the invention can synchronize the coordinate system of each radar to the coordinate system of the vehicle body when a plurality of laser radars are used as the sensors of the intelligent driving vehicle, thereby realizing the alignment operation of the laser radar point cloud on the original data. Meanwhile, the 128-line main laser radar and the two 32-line blind-sweeping lateral laser radars realize the synchronization of the laser radar point clouds to the same vehicle coordinate system, and lay a foundation for subsequent processing such as filtering, downsampling, clustering and identification of the laser radar point clouds.

The specific implementation mode is as follows:

the core problem of the invention is to synchronize the coordinate system of the multi-path laser radar to a unified coordinate system (generally selecting a vehicle body coordinate system), and for the systemThe problem of synchronization of a multi-path laser radar coordinate system can be understood from mathematics that the coordinates of a plurality of objects in different coordinate systems are known, and a rotation matrix and a translation vector in the two coordinate systems are solved. The mathematical formula can be expressed as the following equation: [ x ] of_a,y_a,z_a]^T＝R*[x_b,y_b,z_b]^T+[x₀,y₀,z₀]^TWherein [ x ]_a,y_a,z_a]^TIs a coordinate in a coordinate system A, [ x ]_b,y_b,z_b]^TIs a coordinate in a coordinate system B, R is a rotation matrix, [ x ]₀,y₀,z₀]^TFor translation vectors, directly find R and [ x ]₀,y₀,z₀]^TIt is difficult.

Therefore, the efficient static calibration method for realizing multi-laser radar point cloud alignment based on rviz provided by the invention comprises the following steps:

the first step is as follows: roughly measuring the relative position of each laser radar relative to the coordinate system of the vehicle body, namely a translation vector [ x_a,y_a,z_a]^T；

The second step is that: three calibration cuboids (length 20cm, width 20cm and height 200cm) are respectively placed under coordinates (20m,5m,0m), (30m,0m,0m), (20m, -5m and 0m) in a vehicle body coordinate system to form a triangular potential;

the third step: then writing a topic publishing program, drawing the positions of the three objects in a vehicle body coordinate system according to the actual shapes and the placing positions of the three objects, starting the topic publishing program and rviz, monitoring the point cloud of a multi-path laser radar and the topic of the object by the rviz, and observing the difference between the point cloud imaging and the drawn object position in a three-dimensional view; writing a parameter calibration configuration file, wherein the parameter calibration configuration file comprises six parameters, namely x, y, z, roll, pitch and yaw, and the six parameters respectively represent three parameters of a translation vector, a rolling angle, a pitching angle and a yaw angle.

Wherein the rotation matrix can be further expressed by the following formula:

the fourth step: observing the point cloud imaging movement trend by modifying the angle values of a rolling angle, a pitch angle and a yaw angle until the point cloud imaging movement trend is adjusted to be coincident with three reference objects drawn in rviz;

the fifth step: and finally, fine-tuning the numerical value of the translation vector, namely, the point cloud imaging of the three objects in the fourth step is coincident with three reference objects drawn in rviz, and the point clouds formed by any other objects in the common coverage area of the three laser radars are completely coincident when moving or standing, so that the whole calibration process is completed, and the method is simple and convenient.

The invention can synchronize the coordinate system of each radar to the coordinate system of the vehicle body when a plurality of laser radars are used as the sensors of the intelligent driving vehicle, thereby realizing the alignment operation of the laser radar point cloud on the original data.

The foregoing is only a preferred embodiment of this invention and it should be noted that modifications can be made by those skilled in the art without departing from the principle of the invention and these modifications should also be considered as the protection scope of the invention.