As part of the final exam of a course I was taking, called “Advaned Robot Control and Learning”, we had a five day robot hackathon, which my other team mates and I were fortunate to win.
We were given a Franka Emika Panda robot and an Intel RealSense D435 and the problem setting was to detect and sort trash spread on a “beach”, which consisted of gravel on a desk. The task was to detect different classes of objects and sort them depending on their category (compostable, burnable, cutlery, dishes, cups).
The objects ranged from easy to detect and easy to grasp, such as the red bricks, to hard to detect and hard to grasp objects, such as cigarette buds.
It was a great experience to again work with real hardware in a project that combined different challenges, in computer vision, trajectory planning and robotics control.