Robotics and Autonomous Systems ( ) – Contents lists available at ScienceDirect Robotics and Autonomous Systems journal homepage: www.elsevier.com/locate/robot Implementation and investigation of a robust control algorithm for an unmanned micro-aerial vehicle Arkadiusz Mystkowski ∗ Department of Automatic Control and Robotics, Bialystok University of Technology, Wiejska 45C, 15-351 Bialystok, Poland highlights • New method for implementation and realization of an robust control algorithm. • Hardware-in-the-loop simulations for a micro-UAV. • Consideration of non-linearity, uncertainty, and non-stationarity of UAV’s parameters. • The μ-Synthesis method applied to the UAV’s dynamics control. • The serial connection between the Gumstix micro-computer and the Kestrel autopilot. article info Article history: Received 10 July 2013 Received in revised form 29 March 2014 Accepted 7 April 2014 Available online xxxx Keywords: Micro unmanned aerial vehicle (μUAV) BULLIT μ-Synthesis Robust control Real-time hardware-in-the-loop simulation abstract This paper presents a new method for implementation and realization of an optimal robust control algorithm in the real-time hardware-in-the-loop simulation environment for a mathematical model of the dynamics of the BULLIT micro-aircraft, with consideration of non-linearity, uncertainty, and non- stationarity of its parameters. The robust optimal control method, μ-Synthesis, applied to the autonomous flight dynamics control system of the unmanned aerial vehicle (UAV) meets desired control performances. The serial connection between the Gumstix micro-computer and the Kestrel autopilot extends the ability to implement high order robust controllers. The code of the control algorithm implemented (in the C++ language) in the memory of a Gumstix Verdex Pro single-chip micro-computer enables optimization of the threads-based approach. The hardware-in-the-loop (HIL) simulation mode was implemented in the Kestrel autopilot inner loop, and simulations of all stages of flight were performed in real-time using the actual model of the aircraft and autopilot. Finally, HIL simulations and tests were conducted in order to verify the developed control algorithm. © 2014 Elsevier B.V. All rights reserved. 0. Introduction According to DARPA (Defence Advanced Research Projects Agency), flying objects with overall dimensions up to 15 cm are classified as micro-class and designated with the acronym MAVs (Micro-Aerial Vehicles) [1]. MAV type objects have different aero- dynamic properties than large aircraft, e.g. manned/passenger air- craft, and the approach to the design of the control system is also completely different for MAVs [2–4]. Due to the small control sur- faces of MAV objects and low Reynolds numbers, alternative con- trol methods are being developed [5–7]. Moreover, in the MAVs the interaction between unsteady aerodynamics and structural flexi- bility is critical [8–10]. Therefore, the boundary layer control (BLC) ∗ Tel.: +48 857469230; fax: +48 857469210. E-mail address: a.mystkowski@pb.edu.pl. methods are being used [10–13]. After a micro-aircraft is equipped with on-board electronics for dynamic control in 3D space, navi- gation, and telemetric data transmission, it is able to perform au- tonomous flight missions. Such a micro-aircraft is called a UAV (Unmanned Aerial Vehicle). The subject matter concerning UAVs is the object of many scientific studies. The numerous applications of UAV objects and the requirements that are posed towards them are compiled in the work of the authors: Arning R.K. and Sassen S. [14]. MAV/UAV-controlled systems are unstable, non-linear, and multi-dimensional, with many cross-couplings [9,15]. The param- eters and dynamic properties of a UAV model are non-stationary and variable during flight. The non-stationary nature of system pa- rameters concerns its geometrical model (e.g. change of mass dur- ing flight—fuel consumption, change of the position of the center of mass, deformation of the airfoil, etc.), physical properties, and in particular, its aerodynamic parameters [16,17]. Considering the above, the design of a UAV control system cannot only be based http://dx.doi.org/10.1016/j.robot.2014.04.002 0921-8890/© 2014 Elsevier B.V. All rights reserved.