10.25417/uic.12481025.v1 Matteo Pallomo Matteo Pallomo A Framework for Robotic Grasping and Handover With an Underactuated Three-Fingered Hand University of Illinois at Chicago 2019 Robotics Grasping Handover Baxter 2019-08-01 00:00:00 Thesis https://indigo.uic.edu/articles/thesis/A_Framework_for_Robotic_Grasping_and_Handover_With_an_Underactuated_Three-Fingered_Hand/12481025 Implementing robotic grasp on the Baxter robot without using the default parallel gripper but instead a three fingered underactuated hand. The problem has been solved by using a parallel gripper planner instead of reinventing the wheel for the end-effector in use. This choice allowed to obtain satisfactory results with less design effort and a faster implementation, since it is easier to take advantage of previous works. On the other hand, this approach sacrifice optimality of the grasp, but since the hand is underactuated is not suitable to emulate the human behavior. Moreover, also the handover problem was tackled. We not only wanted to achieve a smooth handover but also pay care to object's safety during the task. In this case we used an optical sensor from a mouse to monitor the object acceleration during the handover. The acceleration feedback is used to re-establish the grasp in those cases where the object was falling too fast.