How to get the detection of checkerboard pattern in camera lidar calibrator app?

Hi,

I have an issue with the calibration of camera and lidar. I would like to know whether the process which I have done is proper or not! But I followed your instructions (Mathworks), but still not getting the detection with the pair.

1. with the help of checkerboard pattern, have captured the data from camera and lidar
2. captured camera data as a frames through mScript.
3. Lidar data captured through the veloview, as a pcap file. From pcap file, pcd file has extracted with the help of matlab convertion. Here is the issue that number of frames in camera and pcd data are different. I couldn't find the similar frames to do the calibration.
4. another issue is, when i upload the lidar data, there won't be the proper ROI to detect the checkerboard. when i tried to adjust the ROI, placing of the ROI box is not happened.
5. Sometimes, 1 frame will get matched but it was detected the different region not the checkerboard pattern.

My obervation from this issues,

1. Due to different frame rate of lidar pcd data extraction from the pcap file, number of frame pair count is varying.
2. Due to the different frame rate issue, couldn't detect the detectected pair.
3. Due to improper ROI, detection is not happened.

Guys, Please let me know what i have missed and where should I do the changes!!!!

I would like to try with cropped pcd data, but I didn't know. If someone helps me to crop the particular region of the input pcd data, will be more helpful to do this analysis.

or

Is any way to capture the sensor data at specified angle? (am using VLP-16 sensor)

Kindly assist me on this...

Thanks in advance.

Answer 1

[Hornett]

Camera and LiDAR calibration can indeed be challenging due to the differences in the nature of the sensors and their data. Based on your description, here are some steps and considerations that might help you troubleshoot the issues:

Synchronization of Data:

• Ensure that the camera and LiDAR data are synchronized. This means that for every frame captured by the camera, there should be a corresponding point cloud captured by the LiDAR at the same time. If the frame rates are different, you may need to interpolate or match timestamps to align the data.

Data Capture:

• When capturing data, make sure the checkerboard pattern is well-illuminated and visible in both the camera and LiDAR data.
• The checkerboard should be placed at various orientations and distances from the sensors to capture a diverse set of calibration data.

Preprocessing of LiDAR Data:

• When converting from pcap to pcd, ensure that you are extracting the relevant frames. You may need to write a script to parse the pcap file and extract the frames that match the camera timestamps.
• If you are having trouble with the ROI in the LiDAR data, you might need to preprocess the point cloud to remove outliers and noise that could be affecting the detection of the checkerboard.

ROI Adjustment:

• If the ROI tool is not working as expected, you may need to manually inspect the point clouds and identify the coordinates of the checkerboard.
• You can then crop the point cloud around these coordinates to focus the calibration process on the checkerboard area.

Calibration Process:

• Please refer to the following documentation for learning more about lidar and camera calibration in MATLAB https://www.mathworks.com/help/lidar/ug/lidar-and-camera-calibration.html

Capture Data at Specific Angles:

• If you are using a Velodyne VLP-16 LiDAR, capturing data at specific angles might require you to filter the point cloud data based on the angle information. The VLP-16 provides azimuth angle information for each point, which you can use to select points within a certain angular range.

I hope this helps