Optimal Control Based Reference Generation for Model Predictive Motion Cueing Algorithms

Abstract

This contribution presents a reference generation method for model predictive control (MPC) based motion cueing algorithms (MCA). The goal of an MCA is to generate a realistic motion feeling for a test driver while keeping the simulator within its workspace limits. To utilize the MPC-based MCA to its full potential, the future reference values over its prediction horizon need to be estimated as good as possible. The approach derived in this contribution models the human driver as an optimal controller in order to predict its future behavior. The numerical solution relies on a real-time conjugate gradient algorithm. Simulation results using a representative driving trajectory demonstrate the performance of the approach as well as its computational efficiency.