ROS / MoveIt / Python / Computer Vision / Intel Realsense / Franka Emika Panda Arm

I worked with a team of 5 to program a Franka Emika Panda arm to make pancakes as our final project for ME 495: Embedded Systems in Robotics. As a team, we each took charge of a sub task and then worked together to integrate all the components into a fully autonomous product. I specifically was in charge of the batter dispensing task, and was heavily involved in developing strategies to increase the success rate of path planning in MoveIt. I utilized single joint control to include universal waypoints for each task and increase the success rate of path planning to tool locations found using computer vision. I additionally implemented action clients to control the gripper and grasp objects at a specified width with a given range of force, an element crucial to the reliability and safety of the batter pouring and flipping tasks. Lastly, I assisted with implementing the necessary transforms to translate coordinates found with our RealSense camera to coordinates in the robot’s frame. Object locations were determined using april tags, and the optimal time to begin the flipping process was determined with canny edge detection measuring the number and rate of change of the contours in the region of interest, much like a human pancake chef looking for bubbles.

https://github.com/algarv/FlipIt-the-Pancake-Flipping-Robot