Regrasping framework for Hybrid Compliant Grippers

Kurzfassung

Robotic manipulation remains among the most challenging problems in the field of robotics. The ability to manipulate objects in a complex environment is crucial for robots to perform tasks autonomously, assisting humans in their daily life operations or reducing the need of human intervention in dangerous environments. When it comes to the choice of the gripper, two main approaches can be identified: the use of complex multi-fingered hands that try to mimic the human hand, or the use of simpler grippers that have a reduced operational space but are easier to control and more robust. The adoption of the first approach is limited by the complexity of the control algorithms and the high cost of the hardware making it unsuitable for wide adoption. In this thesis, we focus on the second approach, and we present a novel manipulation algorithm that can be used to accomplish complex manipulation tasks with a simple hybrid compliant gripper. The adopted gripper combines traditional frictional grasping with suction cup grasping, allowing to create a variety of different grasps depending on the object to be manipulated. The algorithm is evaluated in simulation, and the results demonstrate that the hybrid gripper outperforms traditional unimodal grippers in terms of success rate. Furthermore, the findings highlight that a well-structured planning and control strategy can effectively compensate for hardware simplicity, enabling the execution of sophisticated manipulation tasks without requiring complex or expensive robotic hands.