In order to solve the calibration problem, Livox provides an automatic calibration tool. This technology mainly relies on the isomorphic constraint assumption model of the environment to realize automatic calibration, and only needs to use the original point cloud data of the base LiDAR and target LiDAR. Our tool first mapping the environment by using the Base LiDAR data. Then we use the target LiDAR data to continuously iteratively register to the reconstructed map, relying on the constant assumption of the rigidity of geometric structure to continuously reduce the matching error until the algorithm converges And meets the rigidity invariant characteristic of the calibration matrix (six parallel lines). Finally, a consensus algorithm is used to calculate the final calibration matrix.
Fig.1 Red: Base LiDAR mapping result, Green: Target LiDAR automatic calibration
Cmake, PCL1.7, Eigen
We recommended to install ROS directly, which including the above libraries
sudo apt-get install ros-kinetic-cv-bridge ros-kinetic-tf ros-kinetic-message-filters ros-kinetic-image-transport
mkdir build
cd build
cmake ..
make
This operation will generate three files, namely mapping, calibration, fitline
-
mapping: Mapping Tool
-
calibration: Automatic Calibration Tool
-
fitline: Fit the calibration parameters and calculate the final calibration matrix and Quaternion
-
Put the base LiDAR data into: data/Base_LiDAR_Frames/.pcd (Use 100000.pcd as the first frame file name, and the subsequent frames are accumulated upward)
-
Put the target LiDAR data into: data/Target_LiDAR_Frames/.pcd (Use 100000.pcd as the first frame file name, and the subsequent frames are accumulated upward)
-
Put the initial matrix data into: data/Init_Matrix.txt
Note: The base LiDAR data and the target LiDAR data need to be synchronized in time as much as possible, and the file name and time stamp correspond to synchronization.
Download example data set
Target-LiDAR-Frames
Base_LiDAR_Frames
cd livox_calibration
cp run.sh build/
cd build
sh run.sh
The script first starts the visualize mapping program mapping，Use base LiDAR to build a submap. After that, the script starts calibration program to automatic calibration target LiDAR to base LiDAR，estimated the calibration parameters for each frames. Finally, the script launchs the Curve fitter fitline，to estimate the final parameter matrix.
Fig.2 Out put result of example data set
- Must ensure the data synchronization between sensors
- Must ensure the precision of base LiDAR mapping result
- The movement of the data collection vehicle or platform must be as slow as possible, and we recommended to correct the motion distortion between each frames to ensure the final accuracy
- The accurate initial external parameter matrix is not needed, but it should be roughly aligned to ensure the final accuracy as much as possible
- For Mid-40, Horizon models, the converted PCD file refers to 100ms as one frame
You can get support from Livox with the following methods:
- Send email to [email protected] with a clear description of your problem and your setup
- Github Issues
This code is based on the research below:
@article{gong2018target,
title={A Target-Free Automatic Self-Calibration Approach for Multibeam Laser Scanners},
author={Gong, Zheng and Wen, Chenglu and Wang, Cheng and Li, Jonathan},
journal={IEEE Transactions on Instrumentation and Measurement},
volume={67},
number={1},
pages={238--240},
year={2018},
publisher={IEEE}
}
Developer: Livox
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent