Data-driven Modeling and Optimization of Pinsa Dough Flattening with a Robot

Robotic automation in the food industry is being adopted more and more to satisfy the need for precision and efficiency. The ability to properly manipulate deformable objects plays a key role in food preparation, as ingredients frequently present plasto-elastic behaviors. This paper aims to develop a robotic system able to flatten the dough of a Pinsa Romana. The main elements of the developed system are a robotic arm and a depth camera: the 3D data from this camera is used to assess the state of the dough and determine the trajectories that the robot should follow to process the dough. By adopting a data-driven approach, we developed two algorithms that were able to achieve our objective by using a greedy strategy. These algorithms are then compared in terms of the number of actions needed to achieve the task and the total execution time.