Principal Investigator: Professor Norbert Krüger
Faculty of Engineering – The Maersk Mc-Kinney Moller Institute – SDU Robotics – SDU
In this project, we study simulation and optimization for robotic manipulation of deformable objects (e.g. meat). Simulation of deformable objects requires a substantial amount of computation, and when optimization is placed on top of this hundreds of deformable object simulations are required. Furthermore, to analyze the optimization several optimization runs also have to be evaluated.
Robotic task tends to contain some control parameters, some objective and some uncertainties. Numeric optimization is a powerful tool to pick a set of control parameters that maximize the objective score. But to ensure the uncertainties doesn’t affect the solutions some extra steps must be made. When considering the importance of minimizing the impact of uncertainties, function fitting optimization techniques in particular “RBFopt” are powerful optimization tools.
