A learning approach to the FOM problem

Abstract

Hogan recently provided an heuristic technique called family of modes (FOM) to solve model predictive control (MPC) problems under hybrid constraints and underactuation. The goal of this study is to further develop this new method and to expand its usage in the robotics manipulation community. With that objective in mind, we address some of the method's weaknesses, we provide comparison tools to try to compare the method with traditional MPC solving techniques and we provide a simple and systematic technique to set-up the method's parameters. We conclude the study by presenting our the future lines of research, which consist in generalizing the method for more complex systems and testing it's robustness.