空力モデルに基づくドローンの飛行特性推定

抄録

Robust flight control under adverse environments is required. As an existing method, several systematic control system design methods including H_∞ control have been established. However, as a disadvantage of these methods, control performance may become remarkably conservative, especially when large uncertainty exists. As an element of large uncertainty, modeling error of target dynamics can be mentioned. By calibrating this modeling error, maintainability can be avoided and control performance can be expected to improve. The final objective is to design a controller that can estimate modeling errors in real time and provide adaptive control inputs. As the first step in this paper, we designed an observer that can estimate uncertainty in real time from motion history for modeling error calibration.