Semi-automatic calibration method for multi-line laser radar and visual sensor

Abstract

The invention discloses a semi-automatic calibration method for a multi-line laser radar and a visual sensor. The method is characterized in that the method comprises the following steps: calibrating the internal reference of the visual sensor, inputting a calibration result into a calibration program, and determining a focusing range; moving a calibration plate bottom plate to the front of the visual sensor, then standing still, placing marker standard plates in sequence, and then determining the positions of the marker standard plates through a visual sensor visualization module; recording scale values corresponding to the marker standard boards, inputting the scale values into a prior information calculation module according to a given name and the placement sequence of the marker standard boards, and performing calculation through the prior information calculation module to obtain prior information; and recording data packets of the laser radar and the visual sensor by using rosbag, and when prior information is consistent with configuration information and scale values, starting a calibration program, and outputting an external reference result, an optimizing residual root-mean-square error and a projection evaluation effect. Calibration stability and calibration precision are improved through automatic identification of laser radar feature points and a visual sensor identification verification mechanism.

Description

technical field [0001] The invention relates to the technical field of joint calibration of sensors, in particular to a semi-automatic calibration method for multi-line laser radar and vision sensors. Background technique [0002] Multi-line lidar and vision sensors are the core sensor components in the field of autonomous mobile robots and unmanned driving. Multi-line lidar has a large viewing angle and high-precision three-dimensional space detection performance, and vision sensors have rich visual information and high-resolution , to a certain extent, the two have complementary characteristics. The fusion of multi-sensors has also become a main solution for autonomous mobile robots and unmanned driving to run stably all day and all day long. The joint calibration between multi-sensors is often a multi-sensor It is one of the important basic tasks of fusion, and it is also a hot topic of research at home and abroad. [0003] At present, the joint calibration method of mul...

AI Technical Summary

Problems solved by technology

[0003] At present, the joint calibration method of multi-line laser radar and vision sensor is mainly divided into two directions: online calibration and offline calibration. Dynamic online calibration usually requires a large range and a relatively good calibration environment, but this method is due to vibration, environmental conditions, etc. It is easy to cause unstable calibration results; static offline calibration usually uses a calibration board, which is relatively stable and controllable. There is a way to manually select edge points for line fitting to find corresponding points, but manual selection is easy to introduce human error, and line fitting is due to sparse sampling The problem is easy to enlarge the observation error of the sensor itself, and manually select the corresponding point through the reflection intensity. This method has high requirements for the calibration board. When it is a whole board, it is easy to make it difficult to find enough effective corresponding points.

Images