Which hand eye calibration method is the most popular?

asked 2019-04-25 01:52:29 -0500

Hello, I spent some time looking for Implementations, that solve the Hand-Eye-Problem AX=XB. There are several repositories such as:

Is there a common and most used implementation? Which algorithm is state of the are and offers the best results? Where is its implementation?

Thanks for some hints


edit retag flag offensive close merge delete


have you found out which is the best?

nd gravatar image nd  ( 2019-06-06 08:38:16 -0500 )edit

Apparently the handeye_calib_camodocal package is base on this paper leads to the most robust results.

I would love to have a python implementation of that algroithm, but have not found any implementations yet.

nmelchert gravatar image nmelchert  ( 2019-06-06 09:21:17 -0500 )edit

easy_handeye worked well for me for calibrating the rotation of a Structure Core camera mounted on the wrist of a ur5 robot. However, the translation is off by a few centimeters. Tested over multiple runs and AR markers with different shapes and sizes.

atp gravatar image atp  ( 2019-11-20 10:22:05 -0500 )edit

A few centimeters is a pretty big difference. Your camera calibration could be the cause of this deviation.

nmelchert gravatar image nmelchert  ( 2019-11-20 12:10:22 -0500 )edit

I was just searching information on that and I found that question quite useful! Just to add that if you want to use Ensenso, they have their own implementation in the official API.

Solrac3589 gravatar image Solrac3589  ( 2020-03-02 09:55:17 -0500 )edit

Well, but what kind of algorithm do they use?

nmelchert gravatar image nmelchert  ( 2020-03-02 09:57:14 -0500 )edit

No idea. I am using just the API. I think the backend it's not open source, but I tought was going to work optimally in the case.

Solrac3589 gravatar image Solrac3589  ( 2020-03-09 08:20:45 -0500 )edit