Real-Time Model Predictive Control for Shot Aiming in a Physical Pinball Machine

We examine the shot aiming problem in the context of a physical pinball machine. A parametric switched mode system model is constructed and parameter tuning techniques are presented. A model predictive controller is developed that leverages the tuned model, and experimental results using a physical pinball testbed are examined. The developed controller successfully aimed shots with sufficient precision to satisfy most shot constraints within real-world, commercial pinball machines and provides a good baseline level of performance to use as a point of comparison for future aiming controllers.