Object detection on a rotating platform

 

Tommaso Cavallari, a masters student from the University of Bologna spent his fall internship at Willow Garage working on the recognition and tracking of moving objects on a rotating platform.

He focused on enabling robots to recognize a platform and it's axis of rotation, recognize objects placed on the platform, and understand how those objects rotate over time.

There were a few challenges to overcome during the execution of this task, such as the relative motion between the camera that observes the scene and the objects, obscured objects because of clutter, and altering objects in the scene by other robots or people. 

By the end of his internship Tommaso has been able to develop a solution that can detect one or more objects rotating around a platform. In addition, his solution finds the geometric parameters that describe this movement. In the future, his work will allow robots to figure out how to reliably pick up an object even if it’s moving by either predicting where the object will be located at a specific time or by following the object with the arm while it moves.

Head over github for more information and access to the source code.