Start Date
5-2020 12:00 AM
Description
Real-time model predictive control (MPC) is limited to short time horizons and linear systems because the optimization complexity is too large with long time horizons and nonlinear systems. For this reason, MPC is typically accomplished using linearized models and convex optimization solvers. We seek to explore evolutionary algorithms allowing for nonlinear models and constraints, non-convex costs, and extended time horizons.
Our contributions include extending nonlinear evolutionary MPC to flight vehicles, fixed-wing and multirotor UAVs, as well as enhancing the evolutionary algorithm. We also intend to parameterize the design space of the optimization to reduce solve times. These contributions validate the robust and effective nature of the algorithm.
Included in
Evolutionary Nonlinear Model Predictive Control on a Fixed-Wing and Multirotor
Real-time model predictive control (MPC) is limited to short time horizons and linear systems because the optimization complexity is too large with long time horizons and nonlinear systems. For this reason, MPC is typically accomplished using linearized models and convex optimization solvers. We seek to explore evolutionary algorithms allowing for nonlinear models and constraints, non-convex costs, and extended time horizons.
Our contributions include extending nonlinear evolutionary MPC to flight vehicles, fixed-wing and multirotor UAVs, as well as enhancing the evolutionary algorithm. We also intend to parameterize the design space of the optimization to reduce solve times. These contributions validate the robust and effective nature of the algorithm.
Comments
Due to COVID-19, the Symposium was not able to be held this year. However, papers and posters were still submitted.