High Endurance, Micro Aerial Surveillance and Reconnaisance Robot

Jayant Ratti, Member, IEEE and George Vachtsevanos, Senior Member, IEEE

Abstract— Micro Aerial Vehicles (MAVs) have gained a sig-nificant amount of research lately, with a number of universitiesand industry sponsors paving the way with micro flying robotsto perform Intelligence, Surveillance and Reconnaissance (ISR)Missions. However, much of the work done in flapping wingMAVs till date has not shown performance improvements overtheir VTOL, rotary-wing counterparts. Research and develop-ment over the years has shown that insects and birds haveunmatched flying capabilities in the Low Reynolds NumberRegime. Their phenomenal flight performance is attributed toamong others the highly energy efficient actuation systems &their power to weight ratios, efficient aerodynamics & wing con-figurations, novel & high-speed flight-control and processing.The paper proposes and illustrates the developments thus fartowards the design of a novel, high endurance, flapping wingmicro aerial robot for surveillance and reconnaissance. Theentire design and development efforts, including aerodynamics,actuation-mechanisms, flight controls, embedded controls &processing and finally flight testing is abridged and presentedherein.

I. INTRODUCTION

The applications for MAVs envisioned thus far have been

largely focused on ISR missions performed in swarm or

stand-alone configurations. The MAVs will be tasked with

surveying and patrolling an area, deemed hazardous or un-

navigable by soldiers, armed personnel or humans in general.

The successful deployment and application of such micro

flyers warrants that the MAVs be able to sustain flight for

appreciably long durations. The prospect of such an MAV

flight has so far eluded researchers. The MAV platforms

inspired from flapping wing, biological counterparts have

shown very limited flight endurance [1], [2], [5], [15],

[16]. The majority of flight capable flapping-wing MAVs

have very limited VTOL/hovering capabilities. Their primary

functionality is to mimic fixed-wing flight with the primary

difference of producing forward thrust by the aid of flapping

air backwards. The requirement is to have MAVs be able to

hover and sustain long-flight endurance at the same time. The

simultaneous inclusion of these two primary requirements in

an MAV calls for a paradigm shift in the development of

enabling actuation mechanisms and control techniques.

Manuscript received January 15, 2011. This work was supported by theAir Force Office of Scientific Research under contract No. FA9550-10-C-0036. We acknowledge their support

J.Ratti, Ph.D. Robotics Program, School of Electrical and Com-puter Engineering (phone: 404-312-2644; fax: 404-894-4641; email:jayantratti@gatech.edu).

G.Vachtsevanos, Prof. Emeritus, School of Electrical and ComputerEngineering (email: george.vachtsevanos@ece.gatech.edu).

The authors represent the Intelligent Control Systems Laboratory, Schoolof Electrical and Computer Engineering, Georgia Institute of Technology,Atlanta, GA 30332-0250 USA

TABLE I

DARPA: MAV DEFINITION / DESIGN REQUIREMENTS

Specifications Requirements DetailsSize < 15.24cm (6 in) Maximum dimension

Weight ~100 g Objective GTOWRange 1 to 10 Km Operational Range

Endurance 60 min Loiter time or stationAltitude < 150 m Operational ceilingSpeed 15 m/s Maximum flight time

Payload 20 g Mission dependentCost $ 1500 Maximum cost

II. STATE OF THE ART

Table II1 showcases some of the state of the art MAVs

that have been developed over the years [1], [2], [5], [15],

[16]. At present, a uniformity among most of the MAV

designs can be seen from their propulsion / actuation, a

variation of a simple reciprocating crankshaft mechanism;

symmetric flapping through mechanical coupling to a single

rotary actuator; restricted control schemes and low energy

efficiency attributed to the use of Frequency Modulated

flapping, as opposed to Amplitude Modulation, as used in

insect and bird flight [1], [6], [12], [14]. Elastic storage and

reuse of pectoral muscle energy is not facilitated. Complete

6 DoF dynamics attained by addition of an airplane like tail

rudder or a tail elevator.

A. Trade Study, Preliminary Design Qualification

Limitations of fixed, rotary wing technologies heighten

with size, following the Reynolds number decrease with

smaller wing surfaces2. However , insects and birds are adept

flyers at low Reynolds numbers. A lot of research in the

unsteady Reynolds number region has also supported the

fact[6]. Flight comparison of the flying mechanisms between

a bird, humming bird, butterfly and dragonfly [12] was made

to choose the most suitable MAV design (Fig. 1).

III. QUAD WING MAV DESIGN: BACKGROUND

A. MAV Design Conception: QV Design

Flight mechanism comparison between a bird, humming-

bird, butterfly and a dragonfly was made to choose the

optimal MAV design; a dragonfly-like design [9], [12] was

chosen as the inspiration for the QV3 design was born. Fig. 2

1http://www.ornithopter.org2Aerodynamic effects in the low Reynolds number region are still a

subject of much research [7]3QV - Stands for Quad-Winged MAV. The name has been chosen to

abbreviate the most basic appearance of the proposed Quad Winged MAVDesign

978-1-61284-481-7/11/$26.00 ©2011 IEEE 1

TABLE II

PRESENT STATE OF THE ART 2000-02

2000

MicroBat

2000s

Univ. of Florida

2002

Mentor

2006

Delfly

2007

Nathan Chronister

2009

Petter Muren

Fig. 1. Trade study of natural flyers: Insects and Birds

Fig. 2. Flight Modes for the QV Design

illustrates the in-flight control of primitive maneuvers by the

coordinated control of power distribution to the individual

wings.

B. QV Developmental PlanThe MAV program has been centered around the parallel

research and development scheme of various software /

hardware elements illustrated in Fig. 3 [7], [9], [10], [12],

[13].

IV. ENERGY RESERVE ENHANCEMENT

The QV design’s inherent advantage lies in providing the

MAV with a many fold increase in lift, thus the ability to

carry higher payloads in the form of avionics and battery

packs.

A. Energy Enhancement: Multi-Wing MAVFrom Fig. 5[13], the energy saving per actuator is demon-

strated for multi-wing vehicles with changing avionics power

Fig. 3. Program Objective and Implementation Flowchart

Fig. 4. 2, 4, N-Wing MAVs: Repeating the Basic Unit

consumpton, showing energy savings of the order of 100%

and more in cases. Energy efficiency on the QV design

goes beyond that provided by the four-wing configuration

over a two-wing configuration alone; the inclusion of elas-

tic/restorative wing flapping further improves the energy

efficiency of the MAV. It has been proved [13] that by adding

a passive spring at the wing joints of the MAV, considerable

torque (and in turn energy) savings is observed; which is

greater for lower air-damping on the wings.

V. FIXED FREQUENCY, VARIABLE AMPLITUDE

Definition 5.1: Fixed Frequency, Variable Amplitude(FiFVA) Control Problem: An actuation mechanism

converting a rotary motion to reciprocating motion, by the

use of a crank mechanism exhibits a low efficiency state. The

system amplitude is always constant, and thus the system

can accelerate or decelerate based on how fast the wings are

flapping (Frequency Modulation). But for a fixed frequency

flight system (insects/birds), changing flapping frequency

disturbs the spring-mass system dynamics, resulting in more

energy consumption and poor performance.

2

Fig. 5. Energy Saving / Flight Time Increase Curves

Fig. 6. Kinematic Illustration of the Basic Wing Mechanism

A. Non-Linear State Space Model for Each Wing

The idea of using on the QV, a system which maintains

constant frequency but controls lift and maneuvering by

amplitude modulation, better exhibits the flight dynamics

of insects and birds. Such a system has been attempted but

met with limited success [4], [8]. The mathematical model

of the actuation mechanism has been defined in [9], [12].

The model is based on two coupled mechanisms in series,

a four-bar linkage mechanism and an inverse crankshaft

mechanism. Fig. (6). Our two FiFVA actuators are as follows:

1) Solenoidal - One-Spring Configurations: The

solenoidal / magnetic actuation methodology was designed

to produce FiFVA actuation and control using just one

actuator with only one control/power input. The actuator

was also miniaturized to weigh 5 grams and produced an

output torque of 75gcm. At 3.6 volts, the coil drew 500

milliamps and reached a resonant frequency of 20 Hz [10].

2) Hypocycloidal Gear Train: A hypocycloidal gear train

was examined since by the use of an additional worm gear

arrangement, amplitude modulation is possible, providing

FiFVA control over wing flapping (Fig. 8) [10].

�

Fig. 7. Micro Solenoidal Actuator: CAD (Left), Working Prototype (Right)

Fig. 8. Inter-meshing of the internal gears (Left); Working Prototype(Right)

VI. AERODYNAMICS

A. Grid and Wing Mesh

The unsteady flapping+feathering motion of the cambered

elliptical wing was simulated in Ansys Fluent V12.14. Figure

9 shows the unstructured tetrahedral grid generated using

Gambit5. The overall grid was divided into 2 parts: an inner

smaller spherical domain bounding the wing motion with a

fine mesh to ensure good resolution of the flow around the

wing, and an outer spherical domain with a coarser mesh.

B. Vortex Formation

Fig. 10 represents the formation of leading edge vortex

which plays an important role in production of lift for

unsteady flapping wings at low Reynolds Number. At time

t=0, the wing is horizontal. Due to the increased acceleration

of the wing from its initial position, a leading edge vortex

grows from the base towards the tip. The growth of the

vortex is maximum at a point about 3/4th of wing span.

The formation of this vortex is believed to be related to the

dynamic stall phenomenon due to which a wing produces

high lift even at high angles of attack.

4through a commercially available CFD software package5a commercially-available software for grid generation

Fig. 9. Wing Mesh (Left); Mesh Grain/Boundary (Right)

3

Fig. 10. Leading Edge Vortex Dynamics on Wing Surface (From Left): at 0s; T/8 s; T/4 s; T/2 s; 7T/8 s

Fig. 11. Flapping Mechanism

VII. FLAPPING HARDWARE AND PROTOTYPING

A. Load Cell and Wind Tunnel Rig Designs

Fig. 11 is a scale model of the wing actuation mechanism,

connected to a 6-axis load cell.

B. Feathering Mechanism

Addition of an extra actuator proves bulky and difficult

to install. The feathering mechanisms we have implemented

constitute spring-loaded passive feathering of the wings to

conform to fixed frequency, variable amplitude flapping as

exhibited by birds and insects.

1) Passive-Feathering - Designs: Some of the modular

passive feathering wing designs tested on the flapping setup

shown in Fig. 11 are presented in Fig. 12.

VIII. MICRO ARCHITECTURE AND CONTROL (MARC)

AVIONICS PLATFORM

The conventional Mini and Large scale UAV systems span

anywhere from approximately 12 inches to 12 feet; endowing

them with larger propulsion systems, batteries/fuel-tanks,

which in turn provide ample footprint and power reserves for

on-board avionics and wireless telemetry. The conventional

MAV (Table. I) can have a maximum dimension of 6

inches and weighs no more than 100 grams. Under these

tight constraints, the footprint, weight and power reserves

available to on-board avionics is drastically reduced. The

paper presents the advent of a new line of Micro Architec-

ture and Control (MARC) avionics systems with very low-

power, multi-sensor, multi-processor avionics interconnect

architecture designed specifically with the power and payload

Fig. 12. Passive-Feathering Designs

Fig. 13. Embedded Hardware Architecture

constraints of MAVs in mind towards matching performance

to their larger scale UAV cousins.

A. Avionics Design

Fig. 13 shows the outline of the embedded hardware on-

board the MAV. The selection of the processors for this

embedded device involved many important considerations

such as power consumption, size/weight and performance.

1) Embedded Vision and Wireless Telemetry: The ground

station tele-operates the MAV and/or uploads waypoints for

semi-autonomous flight navigation. The MARC collects in-

ternal state information, environmental data including aerial

4

Fig. 14. Wireless link between the MAV and the ground station

Fig. 15. MARC-1 (dsPIC33F) Single Core

video/images during flight and transmits this data via a

block data packet to the ground station also through the RF

Transceiver; Fig. 14 briefly illustrates the link.

IX. REAL TIME FLIGHT CONTROL

A. Controller Development

The DSP performs all the Flight Control and Navigation

Tasks Required on the MAV. It interfaces with the FPGA

via the EMIF interface to acquire all sensor data and ground

station commands.

The MAV control can be segregated into the flow diagram

(inspired from [3], [17], [18]) as shown in Fig. 17. The

energy controller controls the power delivered to the under-

actuated actuators, in effect controlling the three body angu-

lar rates of the system. The control law used for achieving

the desired energy has been presented in [9], [12].

B. Real Time Operating System Architecture

The architecture of the Hardware in the Loop (HILS) setup

utilizing the RTOS [3] has been shown in Fig. 18. The archi-

tecture has been designed to keep the code flexible enough

to allow adding higher level tasks to mission directives sent

by the ground station. All this can be achieved without

engrossing into the system level intricacies of handling and

programming a micro-controller/microprocessor. The flexi-

bility of the architecture also makes it extremely efficient

to debug faults in low-level, mid-level or high-level tasks,

without having to re-code/interfere with the other tasks.

C. Results: Power Drain and Weight

The goal of the MARC-2 board was to keep all the

avionics within a total weight of 15 grams and the on-board

Fig. 16. MARC-2 (Cyclone III, TI 55xx DSP) Dual Core

Fig. 17. MAV Hierarchical Control Structure

Fig. 18. RTOS Architecture for Flight-Control and Path Planning

avionics power requirement within 0.75 Watts. Towards that

end the different processors and peripherals were selected.

The total weight of the MARC-2 Avionics Suite was 15.75

grams and the total power consumption was 0.65 Watts.

X. MAV DESIGNS BASED ON DIFFERENT ACTUATION

SCHEMES

A. Modular MAV Configurations

At present there are four different MAV prototypes being

constructed in our laboratory conforming to the QV design

specifications; each has four wings to produce Lift & 6DoF

control. Some of the 3D prototypes are shown in Fig. 19.

5

Fig. 19. MAV Designs based on Coil Actuators, Micro Linear Actuators,Cam-Follower Drive Trains

Fig. 20. 3DoF Benchtop Prototype

B. 6DoF Testing of the First MAV Prototype

A Test-Bench Simulator has been designed to perform

3DoF motions and maneuvers. The goal is to test different

control algorithms and their performance on a bench top

setup without the need to initially subject the prototype to

unnecessary wear and tear, crashes and in-flight failures. The

vehicle has be made 6DoF-capable by the installation of an

indigenous, low weight Autopilot [11].

XI. CONCLUSIONS

The paper presented the novel designs & developmental

progress made towards the engineering of a high endurance,

micro aerial robot for ISR operations. The advent of a novel

flight control and MAV configuration was illustrated in the

form of a quad wing MAV (QV). The discussion was led onto

the design and successful development of a new actuation

methodology for MAVs based on the dynamics of insect

flight. The Fixed Frequency, Variable Amplitude (FiFVA) ac-

tuation was engineered and used to successfully demonstrate

the possibility and effectiveness of replicating insect wing

flapping mechanisms. High performance, onboard processing

capability is essential for any surveillance robot and thus

the stringent size, weight and power consumption limitations

of micro flyers was addressed in the form of the MARC

avionics control board. The aerodynamics of the MAV were

researched towards the design and development of efficient

wings and flapping/feathering schemes. The research aims

to improve energy efficiency of MAVs, resulting in higher

endurance micro robotic aerial vehicles, with the capacity

for both military and civilian applications.

ACKNOWLEDGMENT

We thank the Air Force Office of Scientific Research for

sponsoring the research and development of the project. We

also acknowledge the support of our industrial partner Impact

Technologies, LLC. The authors thank the undergraduate stu-

dent members and research scholars for their contributions:

Seong-Joo Kim, Jung-Ho Moon, Emanuel M. Jones, Thomas

D. Pappas, Andrew Punnoose, Aaron T. May, Maha Hosain

and Joshua A. Sandler. We also acknowledge the facilities

and resources provided by Georgia Institute of Technology

for the research.

REFERENCES

[1] Adam Cox, Daniel Monopoli, Dragan Cveticanin, Michael Goldfarb,and Ephrahim Garcia. The development of elastodynamic componentsfor piezoelectrically actuated flapping micro-air vehicles. Journal ofIntelligent Material Systems and Structures, 13(9):611–615, 2002.

[2] J. Grasmeyer and M. Keenon. Development of the black widow microaerial vehicle. In 39th AIAA Aerospace Sciences Meeting and Exhibit,2001.

[3] Dongwon Jung, Jayant Ratti, and Panagiotis Tsiotras. Real-timeimplementation and validation of a new hierarchical path planningscheme of uavs via hardware - in - the - loop simulation. Journal ofIntelligent and Robotic Systems, 54(3):163–181, 2009.

[4] Jishnu Keshavan, Norman M. Wereley, and Alfred Gessow. Designand development of a high frequency biologically-inspired flappingwing mechanism. In 48th AIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and Materials Conference, Honolulu, Hawaii,2007.

[5] R. Kornbluh. Project mentor: Biologically inspired platform. InKeynote Presentation at the 8th AIAA/CEAS Aeroacoustics Confer-ence, June 2002.

[6] Darryll J. Pines and Felipe Bohorquez. Challenges facing future micro-air-vehicle development. Journal of Aircraft, 43(2):290–305, 2006.

[7] Daniel Prosser, Taher Basrai, Jason Dickert, Jayant Ratti, AgamemnonCrassidis, and George Vachtsevanos. Wing kinematics and aerodynam-ics of a hovering flapping micro aerial vehicle. In IEEE Aerospace,2011.

[8] David L. Raney and Eric C. Slominski. Mechanization and controlconcepts for biologically inspired micro air vehicles. Journal ofAircraft, 41(6):1257–1265, 2004.

[9] Jayant Ratti, Rohan Goel, Seon-Joo Kim, Jung-Ho Moon, Thomas D.Pappas, George Vachtsevanos, and Mike Roemer. Bio-inspired microair vehicle: Design and control issues. In AIAA Infotech@Aerospace,2010.

[10] Jayant Ratti, Emanuel M. Jones, and George Vachtsevanos. Fixedfrequency variable amplitude (fifva) actuation systems for microaerial vehicles. In IEEE International Conference on Robotics andAutomation@ICRA, 2011.

[11] Jayant Ratti, Jung-Ho Moon, and George Vachtsevanos. Towards low-power, low-profile avionics architecture and control for micro aerialvehicles. In IEEE Aerospace, 2011.

[12] Jayant Ratti and George Vachtsevanos. A biologically - inspired microaerial vehicle: Sensing, modeling and control strategies. Journal ofIntelligent & Robotic Systems, 60(1):153–178, 2010.

[13] Jayant Ratti and George Vachtsevanos. Towards energy efficiency inmicro hovering air vehicles. In IEEE Aerospace, 2011.

[14] J. M. Rayner. A new approach to animal flight mechanics. Journalof Experimental Biology, 80:17–54, 1979.

[15] P. Samuel, J. Sirohi, L. Rudd, D. Pines, and R. Perel. Design anddevelopment of a micro coaxial rotorcraft. In Proceedings of the AHSVertical Lift Aircraft Design Conference, American Helicopter Society,Alexandria, VA, Jan 2000.

[16] T. Nick Pornsin sirirak, S.W. Lee, H. Nassef, J. Grasmeyer, Y.C. Tai,C.M. Ho, and M. Keenon. Mems wing technology for a battery-powered ornithopter. In Proc. of the 13th IEEE Annual InternationalConference on MEMS, 2000.

[17] G. Vachtsevanos. Handbook of Fuzzy Computation, chapter Hierar-chical Control, pages 42–53. Institute of Physics Publishing, 1998.

[18] G. Vachtsevanos and B. Ludington. Unmanned aerial vehicles:Challenges and technologies for improved autonomy. In WSEASTransactions on Systems, volume 5, pages 2164–2171, 2006.

6