Optimal scheduling for robotized pick and place packaging systems

Production systems come in diverse forms of layouts and configurations, all with the aim of increasing productivity and plant efficiency. As the system grows in size, understanding the interaction of the systems’ parameters and variables becomes far from being an intuitive task. For this reason, system providers often offer software tools to simulate to a high degree of confidence the system dynamics. These tools are easy to use when considering sizing the plant to satisfy different requirements and specifications of that particular plant. However, these tools are complex and time consuming to use when it comes to developing and testing new control algorithms. Robotized Pick-and-Place systems are a subset of production systems and have the same previous considerations. These systems consist of conveying mechanisms and more than one robot to perform the picking and placing task. This requires a carefully designed supervisor that assigns the sequence of pieces and boxes to the team of robots. The aim of this thesis is to develop both online and offline solutions that assure a correct manipulation ofproducts and boxes. The online solution is formalized as Mixed Integer Linear Program (MILP) and applied to the plant as a model predictive controller (MPC) with a receding horizon. The MILP problem has been developed with two lines of thoughts. One follows the idea of solving the assignment problem, and the other solves a network flow problem. Both of these solutions result in a superior solution to the banal approach adopted in many industrial solutions. At this stage, the solutions are computationally heavy to be feasible in an industrial application. However, the results of the optimal solutions have led to developing an offline solution following a simulation based optimization technique. This technique considers a widely-adopted control strategy and enhance it by manipulating a number of plant parameters. The best solution is then applied to the plant. The superiority of this optimization technique lies in the fact that it can easily be applied to different plant sizes and configuration.