Orientation Hackathon: Pen Thief
Computer Vision / Python / PincherX 100 / Intel Realsense
For the final challenge of our hackathon orientation, we used Intel Realsense D435i cameras to locate a pen and direct a Trossen PincherX 100 robot to grab it. This was an individual challenge, and my first experience with computer vision algorithms. I leveraged the purple color of the pen to identify pixels within a color threshold, and the OpenCv and PyRealsense libraries to estimate the position of the pen in 3D space. Then, the algorithm incrementally increased or decreased the angle of the arm’s base joint until it was determined to be within range of the pen, and finally increased the elbow and shoulder joint angles to extend the end effector to reach the depth of the target pixels and close the grabbers. No pens were seriously harmed in this project (though one was dropped many times).
https://github.com/algarv/Pen-Grabber