Visual Servoing for Image with lines

asked 2019-09-09 04:27:36 -0500

Sushil_ros gravatar image

updated 2019-09-09 13:32:52 -0500

jayess gravatar image

Dear ROS community,

I want to ask one question regarding visual servoing in ros,

I have an image with all parameters such as extrinsic and intrinsic parameter so I want to use visual servoing in term of image-based visual servoing, so could you give me some idea so I can do my task,

As prior knowledge, there is some concept to do but I stuck in between

Please help me to figure out this problem

I really appreciate if someone can help me

Thanks a lot


Now I want to use image-based visual servoing to calculate the eular angles, I know there a lot of functionalities in open CV but there is a problem.

  1. How to reduce error in visual servoing?
  2. Is it possible to use IBVS for lanes to calculate the angles such as pitch, yaw & roll?
  3. Is there any opensource platform for visual servoing?
edit retag flag offensive close merge delete


Can you tell us a bit more about what you're trying to achieve, and what you've tried so far? Are you controlling a rover, drone or an arm? what type of visual image are you trying to track?

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2019-09-09 07:12:44 -0500 )edit

So you have a camera and an object you want to pick? Is the orientation of the object varying a lot?

In general, you should have a model of your object. With the help of the model you can estimate the pose of the object you want to grab. Are you stuck there? I can remember, that openCV covers that functionalities.

RosNav gravatar image RosNav  ( 2019-09-09 09:37:31 -0500 )edit

There is an open source library called Visp which provides all the basic functions and examples to perform visual servoing. To compute the angles, if you have a model of the object you are looking at, you'd be able to compute it's pose. In order to reduce the error, there are several things, but the most important are: how much part of the image does your feature cover? (the more, the better). If you need precision with the pose, having more features, usually would lead to a better estimation (in case of occlusion, etc.)

davidllp gravatar image davidllp  ( 2019-09-10 03:35:17 -0500 )edit