Modelling and control of underactuated finger-based gripper with adjustable compliance for grasping of varying and deformable objects

Grasping and manipulation of deformable objects are challenging problems in the food-processing industry. Mainly because objects vary widely in size, mass and stiffness. To prevent damaging the object and to secure a stable grasp, this thesis proposes a finger-based 2DoF per finger tendon-driven gripper design with variable joint stiffness utilizing an agonist-antagonist setup. The gripper approaches an object with high compliance. After an initial grasp, the joints are stiffened to increase load capacity and prevent the object from falling out of the gripper when accelerated. A specific tendon routing configuration is selected to obtain independently controllable joint angles and combined joint stiffness while being underactuated (only three actuators).

Additionally, a standard degree of freedom is implemented to obtain high adaptability for gripper alignment issues and irregular object shapes. This thesis presents a simulation model as a valuable tool for simulation and development. The model is validated using a rapid prototype gripper. The prototype successfully grasped multiple objects of different sizes and stiffnesses. Also, the gripper increased joint stiffness with less than a 5% increase in grasp contact force and less than 0.025 [rad] joint deviation. The prototype showed a joint stiffness range of 0.02-0.48 [Nm/rad].