crigroup / handeye Goto Github PK

Hi,

Could you clarify the units for the rotation_rmse and translation_rmse results returned from the Ros service? My guess is that it's in radians and meters respectively, right?

Also, is there a difference between reprojection error and rmse?

Thanks for the great software. We've seen a lot of interest in MoveIt Calibration, which is built on handeye.

Could you please make a Noetic release?

Hi @fsuarez6, I think this repo is a great work that should be well recognized to wider community. Recently I made an effort to make this repo working as a plugin to MoveIt community. Please see the PR in MoveIt. And I also ported it to ROS2 in my fork. Do you still consider releasing it to Melodic, and welcoming a porting to ROS2 PR? I would be very happy to contribute.

Hi,

Currently working at Niryo, I'm trying some eye-on-hand algorithms in order to see which correspond the most with our expectations which are an overall error under 1 cm (translation error + error generated by orientation error). I'm using a Gazebo simulation in order to compare algorithms result with ground-truths !

I am experimenting issues with your package : I get unsatisfying results. In fac, my translation RMSE is always above 3 cm, and rotation error lead to a ~2 centimeters error on the picking area, and as we use a specific end-effector, we can't deal with these errors

The algorithm also depends highly of how my inputs are sorted : using a random shuffle on the same samples can lead to an translation of 4+ cm.
For instance, on this plot, you can see 30 different iterations (each one corresponding to a shuffle).

• Subplots are X, Y, Z
• Results are sorted according to X
• Red Dotted line corresponds to the mean value of results obtain with handeye
• Blue line corresponds to the expected value

As you can see, results are not very good and kinda noisy

Before going further in this issue, I would like to know if my requirements in precision are achievable with your package ? I didn't find much information about that here or in Danilidis paper

Thanks in advance !

Dear Author

  Hello

 Could you please provide us a tutorial about how to implement the package ?

 Thank you very much.

Hello,
I just read this paper:
https://www.robotic.dlr.de/fileadmin/robotic/stroblk/publications/strobl_2006iros.pdf

Which seems to be quite promising. Do you have an implementation of this?