Calibration error: relative rotation between the lidar and camera is off

[sunemily]

Hi,

We've been using your package to calibrate between a Velodyne VLP16 LiDAR and a ZED camera. We were able to get a relatively good result but our sample Velodyne to ZED projection results seem to be shifted to the left. It seems like the relative rotation between the lidar and camera is off, and were wondering it possible there’s an axis/sign error with the yaw rotation in the transform?

I've attached the config files used in the calibration as well as some sample project result images with LiDAR points shown to be shifted to the left.

Are these types of slight calibration errors expected? If not, what can we do to obtain a better calibration? Thank you!

---

After 100 iterations

Average translation is: 0.0788898 -0.130423 -0.245756 Average rotation is: 0.997837 -0.0464536 -0.0465055 0.0465616 0.998915 0.00124173 0.0463973 -0.00340441 0.998917 Average transformation is: 0.997837 -0.0464536 -0.0465055 0.0788898 0.0465616 0.998915 0.00124173 -0.130423 0.0463973 -0.00340441 0.998917 -0.245756 0 0 0 1 Final rotation is: -0.0457109 -0.997911 0.045659 0.00127881 -0.0457652 -0.998951 0.998954 -0.0456046 0.0033681 Final ypr is: 3.11362 -1.52505 -1.49708 Average RMSE is: 0.0429108 RMSE on average transformation is: 0.0668673

config_file.txt marker_coordinates.txt zed_left_uurmi.txt

[hunkyu]

@sunemily Can you solve it?

[chenwinki]

I have the same problem with you,

-------------------------------------------------------------------- After 100 iterations -------------------------------------------------------------------- Average translation is: -0.136269 0.0661312 -0.128244 Average rotation is: 0.960093 0.0247276 -0.278585 -0.0430343 0.997283 -0.0597897 0.276349 0.0693924 0.958549 Average rotation is: 0.0622946 -0.28232 -0.0257497 Average transformation is: 0.960093 0.0247276 -0.278585 -0.136269 -0.0430343 0.997283 -0.0597897 0.0661312 0.276349 0.0693924 0.958549 -0.128244 0 0 0 1 Final rotation is: -0.0814494 -0.996602 -0.0122436 -0.0645887 0.0175365 -0.997758 0.994582 -0.080476 -0.0657976 Final ypr is: 1.63665 -0.0122439 1.65234 Average RMSE is: 0.0447147 RMSE on average transformation is: 0.0484879

The RMSE is small, but the project result is not good.

[Andyyyyyyyyyyyy]

I have the same problem with you,

-------------------------------------------------------------------- After 100 iterations -------------------------------------------------------------------- Average translation is: -0.136269 0.0661312 -0.128244 Average rotation is: 0.960093 0.0247276 -0.278585 -0.0430343 0.997283 -0.0597897 0.276349 0.0693924 0.958549 Average rotation is: 0.0622946 -0.28232 -0.0257497 Average transformation is: 0.960093 0.0247276 -0.278585 -0.136269 -0.0430343 0.997283 -0.0597897 0.0661312 0.276349 0.0693924 0.958549 -0.128244 0 0 0 1 Final rotation is: -0.0814494 -0.996602 -0.0122436 -0.0645887 0.0175365 -0.997758 0.994582 -0.080476 -0.0657976 Final ypr is: 1.63665 -0.0122439 1.65234 Average RMSE is: 0.0447147 RMSE on average transformation is: 0.0484879

The RMSE is small, but the project result is not good. I have got the rigid-body transformation using this pkg, but I don't how to visualize the lidar-to-camera projection results, can you show me how to project the point cloud to the monocular camera? Any advice is appreciated!

[SinghiDivyanshu]

hi @Andyyyyyyyyyyyy

Are you able to visualize the LIDAR points on camera points. can you help, i'm also got stuck after getting rigid-body transformation.

[Andyyyyyyyyyyyy]

may be this would help youSee fusion.py------------------ Original ------------------From: SinghiDivyanshu [email protected]Date: Wed,Feb 12,2020 3:23 PMTo: ankitdhall/lidar_camera_calibration [email protected]Cc: Andyyyyyyyyyyyy [email protected], Mention [email protected]Subject: Re: [ankitdhall/lidar_camera_calibration] Calibration error: relativerotation between the lidar and camera is off (#72)hi @Andyyyyyyyyyyyy Are you able to visualize the LIDAR points on camera points. can you help, i'm also got stuck after getting rigid-body transformation.

—You are receiving this because you were mentioned.Reply to this email directly, view it on GitHub, or unsubscribe. [ { "@context": "http://schema.org", "@type": "EmailMessage", "potentialAction": { "@type": "ViewAction", "target": "https://github.com/ankitdhall/lidar_camera_calibration/issues/72?email_source=notifications\u0026email_token=AL2IQDVLG72OT4LRMWZFR5LRCOPYPA5CNFSM4GJSHSCKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOELPW3QY#issuecomment-585067971", "url": "https://github.com/ankitdhall/lidar_camera_calibration/issues/72?email_source=notifications\u0026email_token=AL2IQDVLG72OT4LRMWZFR5LRCOPYPA5CNFSM4GJSHSCKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOELPW3QY#issuecomment-585067971", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { "@type": "Organization", "name": "GitHub", "url": "https://github.com" } } ]

[SiddhantNadkarni]

I have the same issue. Was anyone able to resolve this?

[AnthonyLadewig]

Hi @sunemily

How did you project the lidar data onto an image?

Thanks!