Grasping and manipulation of objects remains one of the most challenging problems in robotics. Objects needing to be grasped can vary wildly in size, mass, stiffness, surface and other properties. In contrast, humans from a very young age perform complex grasping and manipulation tasks effortlessly; the human hand is a marvel in its complexity, capability, and robustness.
To achieve satisfactory robustness, most robotic solutions are highly specialized to a particular class of objects where they can perform sufficiently well. However, even within the same object class, inter-object variation can be significant. One such situation occurs in the food processing industry, where natural products such as fruits, vegetables and meats, even of the same type, are highly variable.
The NWO project FlexCRAFT (project website) aims to bring robust perception, world modelling, control, and gripping manipulation technology to the Dutch agri-food sector. The Robotics and Mechatronics group is involved in P4: Gripping and manipulation. Some of the questions that need to be answered are:
- How do we deal with significant variation in size, shape, stiffness, and surface properties?
- What should an effective gripper for such products look like and what sensing does it need?
- How do we control the gripper for robust manipulation?
This MSc project aims to start addressing these questions. Specifically, focus areas are robust/adaptive control strategies that can deal with variations in size, shape, mass, and stiffness.
After a study on the available literature on grasping deformable objects, the student will develop simple models of compliant/deformable objects, with simple grippers driven by electrical motors. These serve as a basis for the development of control strategies that can deal with unknown/incorrect information about object properties.