Parameters for Nonlinear Model Predictive Control in Unmanned Aerial Vehicle Path-Planning Applications

A procedure to find suitable prediction horizons for UAV applications is presented that is based on a novel stability theory for nonlinear model predictive control (NMPC). The procedure allows the reduction of the required length of prediction horizon, which reduces the onboard computational effort. To this end, a modern stability theory is adopted for NMPC. On the basis of this procedure, an analytical method is presented for the derivation of values for the required prediction horizons and suitable stage cost parameters for typical UAV applications. The most important result of this work is that the prediction horizon can be calculated on the basis of the time to go to the target position exclusively, whereas the stage cost parameters depend on the maximum distance to the target and an arbitrary but feasible turn radius of the UAV only. This allows for a relatively simple parametrization in practice for a wide range of UAV path-planning applications.