Content uploaded by Matěj Karásek
Author content
All content in this area was uploaded by Matěj Karásek on Jan 19, 2015
Content may be subject to copyright.
Robotic hummingbird:
Design of a control mechanism for a hovering
flapping wing micro air vehicle
Matěj Karásek
Supervisor: prof. André Preumont
Public PhD defence
Université libre de Bruxelles
École polytechnique de Bruxelles
November 21, 2014
Flight in nature
2
BBC: Life - Birds
Gliding flight Powered (flapping) flight
Birds
www.public-domain-image.com
PterosaursBats
www.fuodense.dk www.deviantart.com
Insects
commons.wikimedia.org
Different evolution paths !
imgarcade.com
ngm.nationalgeographic.com
BBC: Life –Challenges of life
Fish
Squirrels
Reptiles
BBC: Life - Birds
History of manned flight
3
Daedalus & Icarus,
Greek mythology
Images from: www.wikipedia.org, www.fcdly.com, vzducholod.sweb.cz, http://projetcornu.free.fr/
Prisoners “flying”
kites in China
DaVinci flying
machines
1st manned flight,
hot air balloon,
Montgolfiers
559 1490 1783
1st steerable &
powered balloon,
Henri Giffard
1852 1890 1903 1907
1st steerable
glider,
George Cayley
1st powered
flight,
Clément Ader
1st controlled &
powered flight,
Wrights
1804
1st rotary
aircraft flight,
Paul Cornu
Equivalents nature vs aircraft
4
High lift devices Vortex generators
Norberg, 2002
Unmanned Aerial Vehicles
5
Advantages over manned aircraft:
•Less payload smaller, lighter
•No men on-board operation
in risky environments
•Cheaper + cheaper operation
(UAVs, Drones, Remotely Piloted Aircraft)
•First developed in 1990s
•Remote / autonomous operation
•On-board camera + live video link
•Payload: cameras, sensors, weapons
General Atomics MQ-1 Predator
(1994)
15 m, endurance 24 h, range 1100 km
AeroVironment Wasp (2007)
72 cm, endurance 50 min, range 5 km
Proxdynamics Black Hornet (2013)
10 cm, endurance 20 min, range 1.6 km Miniaturization
Micro Air Vehicles
6
Micro Air Vehicles
•Drones restricted in size and weight
•Operation indoors and outdoors
•Hovering capability
Applications:
Potential for flapping wings
Flapping flight
7
Forward
flight
Fast
Hovering
flight
Slow
I. Cohen Group, Cornell University
http://vimeo.com/6362049
25-33 %
67-75 %
50% : 50%
DISCOVERY –Hummingbird Time Warp
http://youtu.be/D8vjYTXgIJw
Hedrick Lab, University of North Carolina
Hovering flapping flight
Wing tip trajectory in hover
Hummingbirds
Images from: www.wikipedia.org, Flickr (Floris van Bruegel, Sergio Quesada)
Bats
8
Damselfly Megaloprepus
19 cm
Bee hummingbird
5 cm, 1.6 g, 80 Hz
Giant hummingbird
21.5 cm, 24 g, 12 Hz
Fruit fly
0.5 cm, 2 mg, 218 Hz
Hawk-moth Manduca Sexta
10 cm, 1.6 g, 26 Hz
Lift enhancing mechanisms
9
Bomphrey et al. 2009
Van den Berg & Ellington, 1997
Lift enhancing mechanisms
Sane, 2003
•Delayed stall of the leading edge vortex
•Kramer effect (Rotational lift)
•Wake capture
•Clap and fling (some insects)
Flow topology
Song et al., 2014
Hummingbirds
10
BBC: Life - Birds
Wing motion control
11
NATURE - Hummingbirds: Magic in the Air
http://www.youtube.com/watch?v=Hrlr45uGapQ
Fliers with four wings
12
E. van Wijk & J. Schaap - Flight Artists project, Wageningen University
http://vimeo.com/66997670, http://vimeo.com/37896598 BBC: Life - Insects
Flapping wing MAVs
13
Nano Hummingbird (2011)
16.5 cm wingspan
19 g
30 Hz
AeroVironment, Inc. + DARPA
Harvard RoboBee (2013)
3.5 cm wingspan
80 mg
120 Hz
Harvard, Chirarattananon et al., 2014
Two wing, tail-less designs
Tail control Four wing designs
18 cm wingspan
20 g, 10-14 Hz
DelFly (2005)
TU Delft,
deWagter et al., 2014
BionicOpter (2013)
63 cm wingspan
175 g, 15-20 Hz
Festo,
Bionic Learning
Network
Nano hummingbird
14
Developed by AeroVironment, Inc. (5 years)
Financed by DARPA (4 million USD)
The only 2 wing tail-less MAV with on board-power
Project goal
15
Wing length (mm)
Mass (mg)
Adapted from: C.H. Greenewalt,
Hummingbirds. Dover, 1990.
Hovering
Nano Hummingbird
16.5 cm wingspan
19 g
30 Hz
www.avinc.com/nano
Harvard RoboBee
3.5 cm wingspan
80 mg
120 Hz
Chirarattananon et al., 2014
Our target
20 cm wingspan
20 g
20 30 Hz
Tail-less flapping wing MAV with a single wing pair
Thesis overview
16
1. Mathematical modelling
2. Hovering flapping flight stability
3. Control and flight simulation
THEORETICAL PART
PRACTICAL PART
4. Flapping mechanism
5. Control mechanism
Goal: Design a mechanism controlling the flight
by wing motion changes
Flapping mechanism
17
Flight muscles, Ilustra Media (http://youtu.be/aFdvkopOmw0)
?
www.faulhaber.com
Flapping mechanism
18
Flapping mechanism
19
Frame + links: 3D printing
DM ABS, resolution 16 µm
Nylon gears DC Motor
Aluminium / steel
rivets
Flapping motion
20
Photron FASTCAM SA3
2000 fps
1/133 x
Flapping amplitude 180°
Right wing Left wing
Mechanism
(theory)
Flapping frequency 25 Hz
Wing design
21
BoPET (Mylar) CFRP Polyester (Icarex)
DISCOVERY –Hummingbird Time Warp
http://youtu.be/D8vjYTXgIJw
Lift measurements
22
Lift
string
Scale
Lift measurements
23
Lift
Moment
Sensor 1 Sensor 2
Lift =Sensor 1 +Sensor 2
Moment = K (Sensor 1 -Sensor 2)
Wing shape evolution
24
Over 70
designs
built and
tested
Mechanism evolution
25
A: m = 5.2 g
(Feb. 2012)
C2: Lift = 6.4 g, m = 5.8 g
(May 2012)
E2: Lift = 9.6 g, m = 7.5 g
(Oct. 2012)
May 2012
December 2012
Take off demonstration: Motor not in the centre Guidance
Mechanism evolution
26
A: m = 5.2 g
(Feb. 2012)
C2: Lift = 6.4 g, m = 5.8 g
(May 2012)
E2: Lift = 9.6 g, m = 7.5 g
(Oct. 2012)
E4: Lift = 16.1 g, m = 10.1 g (Jan. 2014)
TEST BENCH PROTOTYPES
G2: Lift = 9.6 g, m = 9.0 g (Apr. 2013) J2: Lift = 16.0 g, m = 12.5 g (Jan. 2014)
FLIGHT PROTOTYPES
Uncontrolled prototype
27
•Weight: 12.5 g
•Wingspan: 21 cm
•Power: off-board
•Flapping frequency: up to 24 Hz
•Lift: up to 155 mN 16 g
Uncontrolled prototype
28
•Weight: 12.5 g
•Wingspan: 21 cm
•Power: off-board
•Flapping frequency: up to 24 Hz
•Lift: up to 155 mN 16 g
1/8 x
Flight stability needs to be studied first
Flight stability
29
Yaw
Roll
Pitch
Stable flight
=
Maintaining desired attitude
(body orientation)
Jason Paluck
www.flickr.com/photos/jasonpaluck/4744474530
Hovering:
Roll 0
Pitch 0
Yaw Arbitrary
(constant, or changing
with finite rate)
Mathematical model
30
Rigid body dynamics
(6 DOF)
+
quasi-steady aerodynamics Cycle
averaging
1) Vertical dyn. 4) Yaw dynamics2) Pitch dynamics 3) Roll dynamics
Time varying + periodic
2 DOF1 DOF 1 DOF
2 DOF
Always stable Always stable
Longitudinal flight stability
31
small effect
State space
characterized by 3 stability derivatives
Char. Equation (Root locus form)
“gain”
Opposite sign to
Longitudinal flight stability
32
1) Wings below COG 2) Wings close to COG
(very narrow interval of zw)3) Wings above COG
Root locus:
stable unstable, oscillatoryunstable, divergent
Same situation in lateral system
BUT stable for different wing positions
Whole system
unstable
Flight stability in nature
33
Flies: wings above COG
Flies:
wings above COG
and stable…
How to Fly Right –Science Take
(http://youtu.be/QLhOCIdbV7g)
Fly responding to a blast wave
(http://youtu.be/QH091zFHdQ0)
Flight stability in nature
34
How to Fly Right –Science Take
(http://youtu.be/QLhOCIdbV7g)
No halteres unstable
Stable, because of sensors for feedback:
Ristroph
et al., 2013
1) Halteres
(bio-gyroscopes) 2) Ocelli
SPEED
3) Compound
eyes
BBC: Life on Earth (http://youtu.be/ARLTxG2gh3I)
Charles Krebs,
www.photomacrography.net
BBC: Life on Earth
Active stabilization
35
Char. Equation (Root locus form)
State space model
Halteres (bio-gyroscopes)
Feedback on angular rate
1) Wings below COG 2) Wings above COG and sufficient
feedback gain
stable
unstable, divergent
Passive stability
36
No halteres:
passive stability can be
restored by increasing drag
Ristroph et al., 2013
Passive stability
37
So far, only 4 projects demonstrated stable flight
(actively / passively) in two-winged tail-less MAVs
Take-off: 1/13 x
Real time
•Wings below COG !!!
•Off board power (tether)
•BUT sufficient lift reserve for a radio + battery (2 g)
Flight stability - conclusion
38
Wings below COG Wings above COG
Inherent stability
Passive stabilization
Unstable
Not possible
Possible Possible
Possible
Unstable
Active stabilization
with rate feedback
Flight control by wing motion
39
NATURE - Hummingbirds: Magic in the Air
http://www.youtube.com/watch?v=Hrlr45uGapQ
Flight control in nature
40
Roll Pitch Yaw
Conn et al., 2011
Flight control in nature
41
Flight forward
& backward
via pitch
Flight sideways
via roll
N. Boeddeker & J. Zeil,
Australian National University
DISCOVERY –Hummingbird Time Warp
http://youtu.be/D8vjYTXgIJw
Control strategy
42
Direct control Control via pitch / roll
Up/down TurningForward/backward Sideways
Direct control
Pitch dynamics cascade control:
Pitch
dynamics
PI
+-
P
+-
PI
+-
Attitude control
Speed control
4 DOF control:
Control results
43
Little difference between the original and simplified
(cycle-averaged + linearized) model
Prediction of control moment magnitudes
Control mechanism
44
Pitch moment
Lift
side view
L
L
Roll moment
front view
+D
Yaw moment
-D
top view
Control moment generation (2 flapping wings):
2 strategies developed and tested:
•Wing twist modulation via root bar flexing
•Amplitude and offset modulation via joint displacements
Wing twist modulation
45
Lift force control:
According to Keenon et al. 2012
Increased liftNominal liftReduced lift
+=
-
Pitch moment: Roll moment:
Wing twist modulation
46
Manual bar flexing
Roll
moment
control
Pitch
moment
control
SMA driven bar flexing
± 4mm sufficient moments Short stroke, low bandwidth
different actuators needed
Amplitude & offset
modulation
47
Pitch control Roll control
Flapping amplitude and offset can be controlled
by displacing these joints
Roll servo
Left offset
servo
Right offset
servo
32 mm
Amplitude & offset
modulation
48
•Drive: 8 mm brushless DC (5.2 g)
•Control: 3 x micro-servo (6 g total)
•Wingspan: 21 cm
•Total mass: 21.4 g
Amplitude & offset
modulation
49
Hover
Flapping frequency 15 Hz
Photron FASTCAM SA3
2000 fps, shutter 1/10000 s
Yaw
Roll Pitch
Combined commands
50
Amplitude difference
Offset servo [-]
Roll servo [-]
Average offset
Offset servo [-]
Roll servo [-]
Whole workspace: +/-12°
At zero roll: +/-15°
Whole workspace: +/-24°
At zero offset: +/-40°
Flapping frequency 15 Hz
Moment generation
51
Stable system (pendulum), open loop control:
Original aspects
52
Hovering flapping flight stability
•Pitch and roll dynamics can be characterized
by 3 poles each
•The pole configuration depends on the wing
position
•If wings are above the COG, angular rate
feedback stabilizes the system.
Flapping mechanism
•New 2 stage mechanism with symmetric
output has been developed.
•Take off demonstrated.
Control mechanism
•New control mechanism based on flapping
amplitude and offset modulation has been
developed.
Future work
53
Mechanical design
•Control mechanism (efficiency, actuators)
•Weight reduction
•Lift production (wings, gearbox, motor)
Flapping
mechanism
Control
mechanism Wings
Attitude
sensor
(IMU)
Radio
control
On-board
CPU
Battery
Control and avionics
•Attitude sensor
•Micro controller
•Radio
•Motor speed controller
•…
Acknowledgements
54
Internship students
Alexandre Hua
Neda Nourshamsi
Mathieu Dumas
El Habib Damani
Raphael Girault
Malgorzata Sudol
Yassine Loudad
Romaine Hamel
André Preumont
Hummingbird team
Laurent Gelbgras
Yanghai Nan
Mohamed Lalami
Hussein Altartouri
Other ASL members
Foreign partners
Iulian Romanescu (TU Iaşi)
Mihaita Horodinca (TU Iaşi)
Ioan Burda (UBB)
BEAMS department
ULB students
Servane Le Néel
Arnaud Ronse De Craene
Ilias El Makrini
Tristan de Crombrugghe
Lin Jin
Roger Tilmans
Michael Ngoy Kabange
Nicolas Cormond
Beatriz Aldea Pueyo
Hava Özdemir
My family
My friends
My PhD research was supported by
FRIA fellowship (FC 89554).