Time-Optimal Model Predictive Control Using Feasibility Governors

This paper presents a control strategy to perform time-optimal point-to-point motions using the feasibility governor (FG) strategy. This technique consists of two stages. The first stage is a feasibility governor in which an auxiliary reference is calculated inside a feasible set that drives the system toward the reference along the constraint boundaries. The second stage is a time-optimal model predictive control (MPC) formulation with an appropriate structure which is executed in a set where the reference can be reached within the prediction horizon. Feasible and reachable sets that contain system dynamics and constraints information are computed offline, allowing the methodology to run with short prediction horizons. This reduces the computational load in the online execution. An example of a car represented by a bicycle model performing a lateral movement at constant forward velocity is presented to illustrate the controller's performance.