Distributed Coordination and control Experiments on a Multi-UAV Testbed Ellis T King elor of Engineerin The State University of buffalo, 2002 Submitted to the Department of Aeronautics and Astronautics in partial fulfillment of the requirements for the degree of Master of Science in Aeronautics and Astronautics t the MASSACHUSETTS INSTITUTE OF TECHNOLOGY September 2004 C Massachusetts Institute of Technology 2004 Authe Department of Aeronautics and astronautics August 20. 2004 Certified b Associate professor Thesis Supervisor Accepted by Jaime Peraire Professor of aeronautics and astronautics Graduate Stude
c Distributed Coordination and Control Experiments on a MultiUAV Testbed by Ellis T. King Bachelor of Engineering The State University of Buffalo, 2002 Submitted to the Department of Aeronautics and Astronautics in partial fulfillment of the requirements for the degree of Master of Science in Aeronautics and Astronautics at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY September 2004 � Massachusetts Institute of Technology 2004. Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Department of Aeronautics and Astronautics August 20, 2004 Certified by. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jonathan P. How Associate Professor Thesis Supervisor Accepted by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jaime Peraire Professor of Aeronautics and Astronautics Chair, Committee on Graduate Students
Chapter 2 Hardware In the Loop modeling and simulation The hardware-in-the-loop(HWIL) simulation is only useful if it accurately portrays the vehicle dynamics and if the behaviors observed during fight tests can be repli cated on the ground. This chapter focuses on identifying some of the dynamical modes of fight for the 60 ARF Trainer aircraft, and verifying that the HWIL sim- ulations reflect the dynamics expected from the aircraft being employed. Reduced order models for 4 of the 5 dynamical modes are determined for the trainer ARF 60 aircraft using identification techniques on experimental fight data and analytical predictions based on aircraft geometry and aerodynamic data. Section 2.1 describes the simulation settings used to create the hardware-in-the-loop(HWIL) simulations and Section 2.2 details the procedures used to create models of the aircraft dynam ics from data collected during flight tests and hardware-in-the-loop simulations. I Section 2.3, the Cloud Cap autopilot is tuned for the trainer arF 60 aircraft and the closed loop response for several of the modes is measured using the HWIL simulator 2.1 Hardware in the loop simulations 2.1.1 Aircraft simulation model parameters Aerodynamic, inertial and engine calibration information is provided to the Cloud Cap HWIL simulation application to model the aircraft being fown. For simply
Chapter 2 Hardware In the Loop Modeling and Simulation The hardwareinthelo op (HWIL) simulation is only useful if it accurately portrays the vehicle dynamics and if the behaviors observed during flight tests can be replicated on the ground. This chapter focuses on identifying some of the dynamical modes of flight for the 60 ARF Trainer aircraft, and verifying that the HWIL simulations reflect the dynamics expected from the aircraft being employed. Reduced order models for 4 of the 5 dynamical modes are determined for the trainer ARF 60 aircraft using identification techniques on experimental flight data and analytical predictions based on aircraft geometry and aerodynamic data. Section 2.1 describes the simulation settings used to create the hardwareinthelo op (HWIL) simulations, and Section 2.2 details the procedures used to create models of the aircraft dynamics from data collected during flight tests and hardwareinthelo op simulations. In Section 2.3, the Cloud Cap autopilot is tuned for the trainer ARF 60 aircraft and the closed loop response for several of the modes is measured using the HWIL simulator. 2.1 Hardware in the loop simulations 2.1.1 Aircraft simulation Model Parameters Aerodynamic, inertial and engine calibration information is provided to the Cloud Cap HWIL simulation application to model the aircraft being flown. For simply 31
(a)The Clark YH airfoil geometry △△△△A会 deg (b)The Clark YH airfoil lift and drag curves Figure 2-1: The Clark YH airfoil closely resembles the airfoil used on the trainer ARF 60 aircraft and is used to model wing aerodynamics configured aircraft such as the tower trainer 60 ARF used in the testbed, many of the performance characteristics can be obtained using the geometry of the aircraft, such as the data found in Table 2. 1. Detailed descriptions of the surface geometry, wing lift curves, and engine performance curves enable simulations of the aircraft under realistic fight conditions, providing the input parameters are configured accurately For example, the Clark Yh airfoil closely resembles the trainer ARF 60 airfoil and is used to describe the aerodynamic properties of the main wing on the aircraft 25 Some of the data is shown in Figure 2-1. A more detailed description of the simulator input files is given in Ref. 26
(a) The Clark YH airfoil geometry. −5 0 5 10 15 20 25 30 −0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 α [deg] CL CD (b) The Clark YH airfoil lift and drag curves. Figure 21: The Clark YH airfoil closely resembles the airfoil used on the trainer ARF 60 aircraft and is used to model wing aerodynamics. configured aircraft such as the tower trainer 60 ARF used in the testbed, many of the performance characteristics can be obtained using the geometry of the aircraft, such as the data found in Table 2.1. Detailed descriptions of the surface geometry, wing lift curves, and engine performance curves enable simulations of the aircraft under realistic flight conditions, providing the input parameters are configured accurately. For example, the Clark YH airfoil closely resembles the trainer ARF 60 airfoil and is used to describe the aerodynamic properties of the main wing on the aircraft [25]. Some of the data is shown in Figure 21. A more detailed description of the simulator input files is given in Ref. [26]. 32
Table 2.1: Trainer 60 ARF measurements, experimentally determined iner tias as shown in Subsection 2. 1.1. Symbolic notation is borrowed from Ref. 27 Measurement Value Units Symbol (SI) Wing Spa 1.707 b Wing Area 0.5200 S Chord length 0.305 Wing Incidence 1 de Wing dihedral deg Wing Sweep 0.0de Tail area 0.0879m2 Tail Spa 0.606 b t Fin area 0.0324m2 Fin Span 0.216m Fin Offset X 1.143 Fin offset Z 0.120 h Fin Sy deg A Fin volume 0.0189 0.130 Fuselage Length.270 lb Gross mass 5267kg Empty Mass 4.798 ne Roll iner 0.31 kg.m2I Pitch inertia 0.46kg·m2 Yaw inertia* 0.63kg·m22 Aircraft Inertia ex periment The aircraft pitch, roll, and yaw inertias are important parameters for the accurate HWIL simulation of the aircraft dynamics. Fortunately, due to the small scale of the rcraft, experimental measurements can be easily made for each axis of the aircraft. The experimental setup for the roll axis is shown in Figure 2-2. From the aircraft free body diagram, the tension in each cable, T, is ng
Table 2.1: Trainer 60 ARF measurements, experimentally determined inertias as shown in Subsection 2.1.1. Symbolic notation is borrowed from Ref. [27] Measurement Value Units Symbol (SI) Wing Span Wing Area Chord Length Wing Incidence Wing Dihedral Wing Sweep Tail Area Tail Span Tail Offset X Tail Sweep Fin Area Fin Span Fin Offset X Fin Offset Z Fin Sweep Fin Volume Ratio Fuselage CX Area Fuselage Length Gross Mass Empty Mass Roll Inertia* Pitch Inertia* Yaw Inertia* 1.707 0.5200 0.305 1 5 0.0 0.0879 0.606 1.14 9 0.0324 0.216 1.143 0.120 53 0.0189 0.130 1.270 5.267 4.798 0.31 0.46 0.63 m m2 m deg deg deg m2 m m deg m2 m m m deg m2 m kg kg kg m2 · kg m2 · kg m2 · b S c¯ Γ Λ Λ l b St t t t Sf bf lf Λ hf f V¯f I I I m l Sb b m e xx yy zz Aircraft Inertia Experiment The aircraft pitch, roll, and yaw inertias are important parameters for the accurate HWIL simulation of the aircraft dynamics. Fortunately, due to the small scale of the aircraft, experimental measurements can be easily made for each axis of the aircraft. The experimental setup for the roll axis is shown in Figure 22. From the aircraft free body diagram, the tension in each cable, T, is 2T = mg (2.1) 33
L R一 L sin(a)=R sin(g) ng Figure 2-2: Torsional pendulum experimental setup to determine roll axis inertia, Ixc, for the trainer aircraft. The period of oscillation of a roll angle perturbation, is measured to parameterize the aircraft inertia. The angle a is the small angle deviation of the supporting cables from the vertical position. This experiment was also repeated for the pitch and yaw axes to determine Iyy and Izz respectivel For rotational perturbations applied to the airframe, the product of interior angles and distances must be constant 2 where o is the aircraft roll angle perturbation and a is the small angle deviation of the supporting cables from the vertical position. The differential equation describin the motion of the torsional pendulum is governed by a torsional inertia term and the restoring moment due to tension forces Irao+ 2TRsina=0
Figure 22: Torsional pendulum experimental setup to determine roll axis inertia, Ixx, for the trainer aircraft. The period of oscillation of a roll angle perturbation, φ, is measured to parameterize the aircraft inertia. The angle α is the small angle deviation of the supporting cables from the vertical position. This experiment was also repeated for the pitch and yaw axes to determine Iyy and Izz respectively. For rotational perturbations applied to the airframe, the product of interior angles and distances must be constant Rφ = Lα (2.2) where φ is the aircraft roll angle perturbation and α is the small angle deviation of the supporting cables from the vertical position. The differential equation describing the motion of the torsional pendulum is governed by a torsional inertia term and the restoring moment due to tension forces ¨ Ixxφ + 2T R sin α = 0 (2.3) 34