Real time NMPC for fixed wing UAV applications: - Auto descent NMPC implementation on an embedded device -
Translated title
Real time NMPC for fixed wing UAV applications
Authors
Kanellis, Emmanouil ; Vasov, Georgi Valentinov
Term
4. term
Education
Publication year
2016
Submitted on
2016-06-06
Pages
85
Abstract
Fixed-wing unmanned aerial vehicles (UAVs) often need to land in tight spaces, which demands flexible operations. One way to achieve this is a deep-stall auto-descent landing, where the aircraft deliberately enters a controlled stall to slow down and descend steeply. Doing this safely requires a precise control algorithm. In this thesis, the maneuver is formulated as an optimal control problem using a nonlinear flight model of the fixed-wing UAV. The problem is solved with the open-source ACADO Toolkit, and a nonlinear model predictive control (NMPC) approach is implemented so the controller can run on an embedded device. The thesis presents the process of developing and deploying a real-time NMPC algorithm on embedded hardware.
Fastvingede ubemandede luftfartøjer (UAV’er) har brug for stor driftsfleksibilitet og skal ofte kunne lande på meget begrænset plads. En metode er deep-stall auto-descent landing, hvor flyet bevidst bringes i et kontrolleret stall for at sænke farten og gå stejlt ned. At udføre dette sikkert kræver en præcis styringsalgoritme. I denne afhandling beskrives manøvren som et optimalt kontrolproblem baseret på en ikke-lineær flyvemodel for det fastvingede UAV. Problemet løses med det open source optimeringsrammeværk ACADO Toolkit, og en ikke-lineær modelprædiktiv styring (NMPC) implementeres, så styringen kan køre på en indlejret enhed. Afhandlingen præsenterer processen med at udvikle og implementere en realtids-NMPC-algoritme på indlejret hardware.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords
fixed wing UAV ; NMPC ; ACADO ; OCP ; Numerical optimization ; Real time