Transition control of a tail-sitter UAV using recurrent neural networks

Flores, Alejandro; Flores, Gerardo

doi:10.1109/icuas48674.2020.9213919

Cited by 9 publications

(3 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our previous work [6], we investigate the development of a recurrent neural network (RNN) to obtain the required state function for a feedback linearization of the dynamics of the system. In that work, we take into consideration the fact that an RNN has the capability to predict future system's behavior.…”

Section: Problem Settingmentioning

confidence: 99%

See 1 more Smart Citation

Implementation of a neural network for non-linearities estimation in a tail-sitter aircraft

Flores¹,

Flores²

2020

Preprint

View full text Add to dashboard Cite

The control of a tail-sitter aircraft is a challenging task, especially during transition maneuver where the lift and drag forces are highly nonlinear. In this work, we implement a Neural Network (NN) capable of estimate such nonlinearities. Once they are estimated, one can propose a control scheme where these forces can correctly feed-forwarded. Our implementation of the NN has been programmed in C++ on the PX4 Autopilot an open-source autopilot for drones. To ensure that this implementation does not considerably affect the autopilot's performance, the coded NN must be of a light computational load. With the aim to test our approach, we have carried out a series of realistic simulations in the Software in The Loop (SITL) using the PX4 Autopilot. These experiments demonstrate that the implemented NN can be used to estimate the tail-sitter aerodynamic forces, and can be used to improve the control algorithms during all the flight phases of the tail-sitter aircraft: hover, cruise flight, and transition.

show abstract

Section: Problem Settingmentioning

confidence: 99%

“…In our work, we are mainly interested in the translational dynamics of the tail-sitter (system Σ). For that, we assume that the attitude system Ω is stable by the application of an appropriated control τ ; such a controller can be found in our previous work [6].…”

Section: System Modelmentioning

confidence: 99%

Implementation of a neural network for non-linearities estimation in a tail-sitter aircraft

Flores¹,

Flores²

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Li et al 2 proposed a nonlinear robust controller to handle the high nonlinearities, based on a nominal model and a nonlinear disturbance observer to compensate the uncertainties. Flores et al 3 implemented a recurrent neural network for the estimation of highly nonlinear aerodynamic terms during the transition stage. Due to the difficulty of modeling highly nonlinear tailsitters, Barth et al 4 proposed a model-free control algorithm to estimate and counteract the rapidly changing aerodynamic forces and moments with an intelligent feedback structure.…”

Section: Introductionmentioning

confidence: 99%

Wind tunnel tests of a wing at all angles of attack

Smeur

Croon

2022

International Journal of Micro Air Vehicles

View full text Add to dashboard Cite

Tailsitters have complex aerodynamics that make them hard to control throughout the entire flight envelope, especially at very high angle of attack (AoA) and reverse flow conditions. The development of controllers for these vehicles is hampered by the absence of publicly available data on forces and moments experienced in such conditions. In this paper, wind tunnel experiments are performed under different flap deflections and throttle settings at all possible AoA. The dataset is made open access. Our analysis of the data shows for the tested wing, flap deflections greatly affect the lift coefficient and stall occurs at [Formula: see text] AoA as well as [Formula: see text]. Wing-propeller interaction is studied by analyzing the propeller induced force in the axis orthogonal to the thrust axis, which is dependent on AoA, airspeed, flap deflections and thrust in a nonlinear and coupled manner. The influence of inverse flow on the wing is also discussed: The data confirm that when the airflow over the wing is reversed, flap deflections will affect the pitch moment in an opposite way compared to the non-reversed case, but this opposite effect can be avoided by increasing the throttle setting. The data show the exact relationship between flap deflections and forces in this condition. Moreover, it is found that the flap control effectiveness for a wing with or without spinning propellers is usually higher around zero degrees AoA than at [Formula: see text] and it is more effective to change the flaps from [Formula: see text] to [Formula: see text] than from [Formula: see text] to the respective [Formula: see text].

show abstract

Implementation of a Neural Network for Nonlinearities Estimation in a Tail-Sitter Aircraft

Flores

2021

J Intell Robot Syst

View full text Add to dashboard Cite

Transition control of a tail-sitter UAV using recurrent neural networks

Cited by 9 publications

References 26 publications

Implementation of a neural network for non-linearities estimation in a tail-sitter aircraft

Implementation of a neural network for non-linearities estimation in a tail-sitter aircraft

Wind tunnel tests of a wing at all angles of attack

Implementation of a Neural Network for Nonlinearities Estimation in a Tail-Sitter Aircraft

Contact Info

Product

Resources

About