Position estimation on tennis court

asked 2019-10-02 06:54:22 -0600

huawarobot gravatar image

updated 2019-10-02 07:16:01 -0600

I need to estimate the position of a robot on a tennis court. The robot has a camera that can be used to detect the court-lines and trying to map them to an internal "tennis-court reference map".

My questions are:

-) Is this a good idea and how can it be easily realized with opencv? I am thinking in first unwarping the camera image to get a birds-eye view. Then filtering the white lines and trying to match them in size and position to the reference map. But how exactly can this matching be done with openCV?

-) Are there other, better suitable methods for that application? LIDAR is a problem since the net is not detected reliably and the dimensions of the court are too big.


edit retag flag offensive close merge delete


yea, start with the unwarping (perspective correction). take a course in computer-vision. then try again.

berak gravatar imageberak ( 2019-10-02 07:25:10 -0600 )edit

GPS RTK could be another solution.

Eduardo gravatar imageEduardo ( 2019-10-02 11:12:45 -0600 )edit

GPS would indeed be an idea, however it's supposed to be working indoors so GPS is no option.

huawarobot gravatar imagehuawarobot ( 2019-10-03 00:54:49 -0600 )edit

I am trying to solve the same problem. I would love to get in touch and chat about the problem. Contact me at service@georgebuild.com or at georgebuilt.com

georgeman93 gravatar imagegeorgeman93 ( 2020-04-23 01:31:41 -0600 )edit