ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Picking Up moving object from a Belt

asked 2019-01-04 04:02:43 -0500

Buldoza gravatar image

Hello, im a newbie here for my thesis, I have a 6DOF Cobot already working well with ROS and Moveit , and i want to pick up objects from a moving belt at normal speed. The Challenge here is there is no way to integrate an encoder to the Cobot to send information to the Cobot. So i need to figure a way out on how to implement this.

Since there is no encoder, the idea is to take snapshots of the object while it is moving and then take the data from the Camera and use it to predict the future rendez vous point where the Cobot will pick up the moving object.

For a start the belt will be moving at normal speed. I will like to ask if there are any suggestions about how to go about this? or better recommendations, Examples, books/information will be gladly appreciated.

edit retag flag offensive close merge delete


There may be a ridiculously simple method of adding an encoder onto the belt if you want. In almost all cases placing an optical mouse at a fixed point on the edge of (or underneath) the belt will allow you to use that to determine its speed very accurately.

PeteBlackerThe3rd gravatar image PeteBlackerThe3rd  ( 2019-01-04 10:17:51 -0500 )edit

Hello, The problem is the connection between the belt and the Cobot. The Cobot does not support this connection! there is no possibility to do this sir. Thanks for your input.

Buldoza gravatar image Buldoza  ( 2019-01-07 02:52:34 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2019-01-04 06:11:23 -0500

juanlu gravatar image

updated 2019-01-04 06:11:42 -0500

Since you don't have encoders and will have to rely completely on the camera to determine the future position of the object I foresee one issue to be extrinsic camera calibration, that is knowing where the camera and the robot exactly are in relation to one another or to a third frame. For that I can recommend you to take a look at the industrial extrinsic calibration package.

Another issue I can imagine is grasping which can be more or less complex depending on the geometry and position of the incoming object (for example if it is LEGO pieces it might be ok with having one or two predefined grasps but if it is a cup of tea that can come upside down then it is necessary to both increase processing and maybe number of sensors). You can start by taking a look at this.

Good luck, sounds like an awesome project!

edit flag offensive delete link more


Thanks very much for your insight on this sir, i will have a look at this and will probably be going down this road. The pieces are just small pieces the gripper is 30mm.

Buldoza gravatar image Buldoza  ( 2019-01-07 02:42:18 -0500 )edit

Question Tools



Asked: 2019-01-04 04:02:43 -0500

Seen: 614 times

Last updated: Jan 04 '19