A Hardware-In-loop Platform for Rotary-Wing Unmanned Aerial Vehicles

7/23/2019 A Hardware-In-loop Platform for Rotary-Wing Unmanned Aerial Vehicles

1/12

A Hardware-in-loop Platform for Rotary-Wing Unmanned

Aerial Vehicles

Igor Henrique Beloti Pizetta1, Alexandre Santos Brandao2, and Mario Sarcinelli-Filho3

Abstract This work describes the development ofa platform to deal with simulated and real auto-nomous flights of rotary-wing aircrafts. Such plat-form (called AuRoRa Platform) was designed for usewith commercial miniature helicopters and embedsthe instrumentation necessary to autonomously guidethe aircraft. Aiming at such objective, an electronicboard specific to actuate over the servomotors ofthe helicopter and to get the data delivered by thesensors onboard the aircraft (all them componentsof the shelf), called AuRoRa Boardwas designed andmanufactured. The AuRoRa Platform was implemen-ted to exchange information with such a board, asa high level Hardware-in-the-Loop platform, capable

of running simulations and real experiments, in thelast case actuating over the servomotors of the air-craft through the electronic board aforementioned.Using the AuRoRa Platformone can choose amongstvarious aircrafts (we have already tested it withthe quadrotor AR.Drone Parrot and the miniaturehelicopters ALIGN T-REX 450 and T-REX 600 -these two latter ones are the aircrafts dealt with inthis work) and amongst different controllers as well.The AuRoRa Platform also presents a decentralizedcharacteristic: to avoid overloading a single computerwith the synthesis of the control signals and the onlineexhibition of the flight data, a feature enabled/disbledby the user, it is possible to run the online exhibitionof the flight data in a second computer, using an UDP

communication channel.

I. Introduction

Over recent years there has been a growing interestin unmanned aerial vehicles (UAVs), associated to theincreased processing capacity of micro controlled sys-tems, leading to smaller embedded systems. The mostremarkable dissemination in the scientific community isrelated to UAV rotorcrafts (helicopter and quad-rotor)due to their ability to take off and land vertically, tohover with the possibility of changing its orientation,to move ahead or laterally while keeping the height, tocompletely change their direction of flight and to stop

*This work was supported by CNPq - a Brazilian Agency sup-porting the scientific and technological development, and FAPES -an agency of the State of Esprito Santo supporting the scientificand technological development.

1I. H. B. Pizetta is with the Department of Mechanics,Federal Institute of Esprito Santo, Aracruz - ES, [email protected]

2A. S. Brandao is with the Department of Electrical Enginee-ring, Federal University of Vicosa, Vicosa - MG, Brazil [email protected]

3M. Sarcinelli-Filho is with the Graduate Program on ElectricalEngineering, Federal University of Esprito Santo, Vitoria - ES,Brazil [email protected]

their movement abruptly. Despite these considerable ad-vantages, in comparison with fixed-wing UAVs (airplanesand gliders) or even with lighter-than-air UAVs (balloonsand airships), rotary-wing UAVs are machines extremelycomplex to control, because of the inherent difficulty togenerate propulsion and simultaneously move itself [1],[2].

However, their high mobility makes the rotary-wing UAVs very useful machines to accomplish three-dimensional tasks in places of difficult access and highrisk, or even over large areas. Amongst such tasks, onecan cite those related to public safety (supervision of airspace and urban traffic), management of natural hazards(like volcanoes), environmental management (measure-ment of air pollution and monitoring of forests), interven-tion in hostile environments, maintenance of infrastruc-ture (transmission lines, fluid or gas pipes), and precisionagriculture. In such tasks, the UAVs are more efficientand their use is more advantageous, in comparison withunmanned ground vehicles, for instance, because of thevertical maneuverability [3], [4].

The aforementioned features have motivated manyresearchers from several areas to propose models andflight controllers capable of guiding UAVs in several and

diverse applications. However, from the point of viewof control, these aircrafts are inherently unstable, non-linear, multi-variable, with complex and highly coupleddynamics. Due to these characteristics, an experimentaltest of UAVs generates a great risk, not only for theequipment under development, but also to people nearbyit (after taking off, such vehicles can fly with great speedand their blades can reach high velocities). Such issueshave motivated the creation of quite complex simulators,mimicking the reality quite well, which are of extremeimportance in the development of UAV flight controlsystems.

According to [5], two approaches can be adopted to

develop a simulator: the software-in-loop (SIL) one, inwhich all components are simulated, i.e., sensors, actu-ators, aircraft model and so on, and the hardware-in-loop (HIL) one, in which just some parts are simulated,whereas other parts correspond to real pieces that areeffectively operating acting on the aircraft (e.g., sensorsor even actuators).

Simulation reduces the time spent to develop a te-chnology. In other words, one can perform tests in afaster and less tiring way than running real experiments.Using HIL and simulations, one can validate a system

2014 International Conference on Unmanned Aircraft Systems (ICUAS)May 27-30, 2014. Orlando, FL, USA

978-1-4799-2376-2/14/$31.00 2014 IEEE 1146


2/12

prior to an experimental test, and develop hardware andsoftware that are compatible, avoiding injuring peopleand damaging equipment. In particular, HIL simulatorsare those whose simulation loop deals with the repre-sentation of the environment in which the hardware willwork. Thus, after executing tests with a sufficient numberof environmental conditions and analyzing the behaviorof the hardware, a comprehensive assessment about the

system can be done. Notice that it is the last step beforea real experiment, since one should have the hardwareto simulate the environment. Such approach is presentedin [6], [7], [8]. In contrast, the SIL simulation does notdepend directly on any hardware piece, since it is beingdeveloped to compose the onboard system. Examples ofsimulators using this approach can be found in [9], [10],[11].

In such context, this paper presents the developmentof a HIL platform, capable of simulating and testingUAVs with different mathematical models and controlstrategies, developed starting from smaller versions usedin other works of our research group [12], [13], [14].To give details about such development, the paper ishereinafter split in some sections. Section II presents anoverview of the aircraft model, emphasizing the controlvariables and the degrees of freedom available for control-ling. In the sequel, Section III gives a description of themain sensors available onboard the aircraft, emphasizinga specific board designed and manufactured to be theinterface between the controller and the sensors/actua-tors of the aircraft. Following, Section IV describes thestructure of the platform here implemented, emphasizingits distributed implementation, and Section V showssome examples of simulated and real flights managed by

the proposed platform. Finally, Section VI highlights themain conclusions of the work.

II. Miniature Helicopter Model

To design a controller to guide the autonomous flight ofan aircraft it is necessary to have its model. Although notbeing the main objective of this work, the model of thehelicopter is required to understand how the proposedplatform operates. For more details on this model, theinterested reader should access [14] and [15], for instance.

The UAV is a solid body subject to external forcesand torques, generated by its propellers. Therefore, it is

a rigid body moving in the three-dimensional space. Twoapproaches can be adopted to describe it: Newton-Eulerequations [16], [17], [18] or Euler-Lagrange equations [19],[20], [15]. Both approaches lead to the dynamic model ofthe rigid body. The main difference between them is thereference system: the Newton-Euler approach considersthe reference system whose origin is the center of massof the aircraft, whereas the Euler-Lagrange approachadopts the inertial frame as reference system.

The model adopted in the platform here describedwas developed using the Euler-Lagrange approach (other

models could be adopted without any problem), and isdescribed by [15]

M(q)q+C(q,q)q+F(q) +G(q) =+D, (1)

where q is the vector containing the attitude of theaircraft, M is the matrix of inertia, C is the matrix ofcentripetal forces and Coriolis, F is the friction vector,

G is the vector of gravitational forces, is the vectorwith the control signals applied to the vehicle and D isa vector of disturbances.

ucoluped

ulatulon

MRTR

B

A

TMTTTR

1c1s

f

t

ActuatorDynamics

Rotary WingDynamics

Torque and

ForcesGeneration

Rigid BodyDynamics

Fig. 1. Block representation of the Helicopter dynamic model.

According to [21], [22], [23], the whole model of aminiature helicopter is represented by four interconnec-ted dynamic subsystems, shown in Figure 1, which arethe actuator dynamics (responsible for transforming the

joystick- or computer generated - commands in servomo-tor actuation), the rotary-wing dynamics (witch embedsthe aerodynamic parameters and the thrust generationassociated to the main and tail rotors), the forces andtorques generation process (where occurs the thrust de-composition), and the rigid-body dynamics(which definesthe rotorcraft displacement in the free space).

The block entitled Actuator Dynamics is where thecommands are received and transformed in actuationcommands for the servo motors. This can be remotelycontrolled by a human operator, or by a computer sys-tem onboard the aircraft or on the ground, in the case

of autonomous flight. The second block, Rotary WingDynamics, relates the aerodynamic parameters and thepropulsion generation associated with the tail and mainengines. The following block, Torque and Forces Ge-neration, decomposes the propulsion vectors, accordingthe current orientation and geometry of the aircraft,and converts them into forces and torques applied tothe vehicle. Finally, the Rigid Body Dynamics definesthe displacement and the orientation of the aircraft inthe three-dimensional space as a function of the appliedforces and torques.

Therefore, the first three blocks shown in Figure 1are responsible for receiving control signals and gene-

rating forces and torques which will effectively act onthe aircraft. Such blocks correspond to what is hereconsidered as low-level model. The last block receivesabstract control signals, the forces and torques acting onthe aircraft body, which will cause the movement of therotorcraft in the 3-D space. Such block corresponds to thehigh-level model. In this work, however, just the low-levelmodel is considered, since the objective is to describehow and why certain components of the shelf(COTS)are integrated to the aircraft, in connection with thedeveloped platform.

1147


3/12

As for the high-level model of the aircraft, it will notbe directly addressed in this work, as well as the RotaryWings Dynamics and Torques and Forces Generation.Actually, it is included in the complete model adoptedto design the flight controller (discussed in [15]) usedto test the implemented platform. Details on such topic,however, are out of the scope of this paper.

A. Principle of Operation of the Miniature HelicopterTo describe the helicopter dynamics, the notation and

reference systems are now defined, for better understan-ding. The subscripte represents the terrestrial referencesystem (the inertial one), the subscript h refers to thereference system of the helicopter, centered at its centerof mass, and the subscript s is associated to the spatialreference system, which is the h system rotated to havethe same orientation ofe (see Figure 2).

The attitude of the aircraft correspond to the positionvariables he (the subscript he indicates the helicopterreference system relative to the inertial frame), and theorientation variables he, relative to its own reference

system. Therefore, one has that he= [x y z]3

repre-sents the longitudinal, lateral and normal displacements,whereashe= [ ]

3 corresponds to the roll, pitchand yaw angles.

In order to establish the speeds, one only has to derivethe position coordinates, resulting in se= [ x y z]

3.Deriving again one has the the accelerations se= [ x y z]3. Notice that the reference frame is the spatial one (s)related to the terrestrial one (e). The same procedure isadopted for the angles, resulting in the angular velocitiessh= [ ]

3 and the angular accelerations sh=[ ] 3. The reference relationship is now betweenthe spatial reference system and the reference system of

the helicopter, both with origin at the center of gravityof the aircraft.

To perform the control of these variables, the operatorhas four input commands [22], namely

Airelon: (ulat), which controls the main rotor lateralcyclic pitch (see Figure 3(a)) which produces the rollmovement, resulting in the lateral displacement of thehelicopter;

Elevator: (ulon), which controls the main rotor longi-tudinal cyclic pitch (see Figure 3(b)), which causes the

zh

yh

e

xhye

xe

ze

>

Documents

A Hardware-In-loop Platform for Rotary-Wing Unmanned Aerial Vehicles