Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Smart Control Algorithm for 2-DOF Helicopter
Glenn Janiak Kenneth Vonckx Advisor: Dr. Suruz Miah
Department of Electrical and Computer EngineeringBradley University
1501 W. Bradley AvenuePeoria, IL, 61625, USA
Thursday, November 29, 2018
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 1 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 2 / 36
http://www.bradley.edu
Introduction
Helicopter are important for short-distance travel
air-sea rescuefire fightingtraffic controltourism
Purpose of control system
resistance to turbulenceenable use of mobile device
Which is better?
Fundamental (LQR)Noise Filtering (LQG)Machine Learning (ADP)
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 3 / 36
http://www.bradley.edu
Introduction
Helicopter 1
Helicopter 2
Helicopter N
Proposed
smart control
algorithm of
helicopters
Mobile
device 1
Mobile
device 2
Mobile
device N
Wireless
network
Figure 1: General High-Level System Architecture
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 4 / 36
http://www.bradley.edu
Introduction
This project will:
use a pair of 2-DOF (2-degrees-of-freedom) testing platformsimplement control algorithms on embedded systemuse mobile device for user controlencourage researchserve as an educational tool
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 5 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 6 / 36
http://www.bradley.edu
Background StudyControl Techniques
Various control techniques have been proposed for 2-DOF helicopters suchas:
Sliding mode control [1]
Fuzzy Logic control [2] [3] [4]
Data-driven Adaptive Optimal Output-feedback control [5]
Decentralized discrete-time neural control [6]
These control techniques employ advanced mathematics that are difficultto implement on embedded systems.
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 7 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 8 / 36
http://www.bradley.edu
Background StudyModeling a 2-DOF Helicopter
Figure 2: Model of a 2-DOF Helicopter
Figure 3: Quanser Aero
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 9 / 36
http://www.bradley.edu
Background StudyModeling a 2-DOF Helicopter
Characterized by fixed baseCan change 2 of 3 possible orientations...
Pitch (θ)Yaw (ψ)Not Roll
and cannot change position
x directiony directionz direction
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 10 / 36
http://www.bradley.edu
Background StudyModeling a 2-DOF Helicopter
Motors are attached to the propellers to create thrust due to airresistance
Main - changes pitch angleTail - changes yaw angle
Torque due to rotation also creates a force on opposite axes
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 11 / 36
http://www.bradley.edu
Background StudyModeling a 2-DOF Helicopter
ẋ(t) = Ax(t) + Bu(t), where (1)
A =
0 0 1 00 0 0 1
−KspJp 0 −DpJp
0
0 0 0 −DyJy
and B =
0 00 0KppJp
KpyJp
KypJy
KyyJy
,
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 12 / 36
http://www.bradley.edu
Background StudyModeling a 2-DOF Helicopter
Ksp - being the stiffness of the axes
Kpp - pitch motor thrust constant
Kpy - thrust constant acting on the pitch angle from the yaw motor
Kyp - thrust constant acting on the yaw angle from the pitch motor
Kyy - yaw motor thrust constant
Jp - moment of inertia about pitch axis
Jy - moment of inertia about yaw axis
Dp - viscous damping of the pitch axis
Dy - viscous damping of the yaw axis
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 13 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 14 / 36
http://www.bradley.edu
Background StudyPrior Work
extensive modeling & simulations
implementation of two motion control algorithms (LQR & ADP)
one helicopter
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 15 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 16 / 36
http://www.bradley.edu
Subsystem Level Functional RequirementsBlock Diagram
Application
Transport
Link
Internet
Application
Transport
Link
Internet
Mobile Device Raspberry Pi
Data
Data
Data
Data
UDP
UDP
UDP
IP
IPMAC
Controller
(packet sent through air)
Data
DataUDP
DataUDPIP
DataUDPIPMAC
Figure 4: Communication Model
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 17 / 36
http://www.bradley.edu
Subsystem Level Functional RequirementsBlock Diagram
Select
algorithm
Actual pitch
Desired pitch
Desired yaw
Actual yaw
Pitch motor
Yaw motor
Pitch encoder
Yaw encoder
LQG
LQR
ADP
Figure 5: Low Level Smart Control Diagram
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 18 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 19 / 36
http://www.bradley.edu
Preliminary WorkLQR Simulation
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [s]
0
5
10
15
20
25
30
35
40
45
50
Po
sitio
n [
de
g]
Position
θ
ψ
θd
ψd
(a)
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [s]
-20
-15
-10
-5
0
5
10
15
20
Inp
ut
Vo
lta
ge
[V
]
Voltages
Vp(t) [V]
Vy(t) [V]
(b)
Figure 6: LQR Simulation (a) Position and (b) Voltage w/ Constant Signal
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 20 / 36
http://www.bradley.edu
Preliminary WorkLQR Simulation
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [s]
-100
-50
0
50
Err
or
[de
g]
Error
eθ [deg]eψ [deg]eθ̇[deg/s]
eψ̇[deg/s]
Figure 7: LQR Simulation Error w/ Constant Signal
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 21 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 22 / 36
http://www.bradley.edu
Preliminary WorkLQR via USB
0 5 10 15 20 25 30
Time [s]
-15
-10
-5
0
5
10
15
Pitch
[D
eg
]
Pitch Encoder
(a)
0 5 10 15 20 25 30
Time [s]
-20
-15
-10
-5
0
5
10
15
20
Vo
lta
ge
[V
]
Pitch Voltage
(b)
Figure 8: LQR Pitch Position (a) and Voltage (b) on PC with Square Wave Input
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 23 / 36
http://www.bradley.edu
Preliminary WorkLQR via USB
0 5 10 15 20 25 30
Time [s]
-80
-60
-40
-20
0
20
40
60
Ya
w [
De
g]
Yaw Encoder
(a)
0 5 10 15 20 25 30
Time [s]
-20
-15
-10
-5
0
5
10
15
20
Vo
lta
ge
[V
]
Yaw Voltage
(b)
Figure 9: LQR Yaw Position (a) and Voltage (b) on PC with Square Wave Input
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 24 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 25 / 36
http://www.bradley.edu
Preliminary WorkLQR via Wireless
0 5 10 15 20 25 30
Time [s]
-60
-40
-20
0
20
40
60
Po
sitio
n [
de
g]
(a)
0 5 10 15 20 25 30
Time [s]
-20
-15
-10
-5
0
5
10
15
20
Vo
lta
ge
[V
](b)
Figure 10: Performance in following user’s command (a) tracking pitchangle, and (b) pitch motor input voltage
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 26 / 36
http://www.bradley.edu
Preliminary WorkLQR via Wireless
0 5 10 15 20 25 30
Time [s]
-150
-100
-50
0
50
100
150
Po
sitio
n [
de
g]
(a)
0 5 10 15 20 25 30
Time [s]
-20
-15
-10
-5
0
5
10
15
20
Vo
lta
ge
[V
](b)
Figure 11: Performance in following user’s command (a) tracking yawangle, and (b) yaw motor input voltage.
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 27 / 36
http://www.bradley.edu
Outline
1 Introduction
2 Background StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
3 Subsystem Level Functional RequirementsBlock Diagram
4 Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
5 Parts List
6 Schedule for Completion
7 Future Directions
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 28 / 36
http://www.bradley.edu
Demonstration
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 29 / 36
http://www.bradley.edu
Parts List
HardwareTwo Quanser Aeros
Q-flex2 Embedded Panel
Two Single Board Computers (Raspberry Pi 3 Model B)Android Smart-phone or Tablet(Note that Apple devices could also be used, however modifications areneeded)
SoftwareMATLAB & Simulink
Raspberry Pi Support PackageAndroid Support Package
Quanser Real-Time Control (QUARC)
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 30 / 36
http://www.bradley.edu
DeliverablesSchedule for Completion
2018
Sep Oct Nov Dec
66%Research100%Last Years Work
100%LQG
0%ADP
66%Simulation100%LQR
100%LQG
0%ADP
50%Implementation100%LQR USB
100%LQR Raspberry Pi
100%LQR Android
0%LQG USB
0%LQG Raspberry Pi
0%LQG Android
Figure 12: Gantt chart for Fall 2018
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 31 / 36
http://www.bradley.edu
DeliverablesSchedule for Completion
2019
Jan Feb Mar Apr May
0%Implementation0%ADP USB
0%ADP Raspberry Pi
0%ADP Android
0%Testing
Figure 13: Gantt Chart for Spring 2019
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 32 / 36
http://www.bradley.edu
Future Directions
Two more motion control algorithms (LQG & ADP)
Test plan
Implmentation on 6-DOF Helicopter
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 33 / 36
http://www.bradley.edu
Summary
Embedded implementation of control algorithms
Mobile interface
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 34 / 36
http://www.bradley.edu
For Further Reading I
[1] S. I. K. Q. Ahmed, A.I.Bhatti, “2-sliding mode based robust controlfor 2-dof helicopter,” 11th International Workshop on VariableStructure Systems, June 2010.
[2] W. Chang, J. Moon, and H. Lee, “Fuzzy model-based output-trackingcontrol for 2 degree-of-freedom helicopter,” Journal of ElectricalEngineering Technology, vol. 12.00, no. 1, pp. 1921–1928, 2017,quanser product(s): 2 DOF Helicopter. [Online]. Available:http://www.jeet.or.kr/LTKPSWeb/uploadfiles/be/201705/290520170957344003750.pdf
[3] E. Kayacan and M. Khanesar, “Recurrent interval type-2 fuzzy controlof 2-dof helicopter with finite time training algorithm,” inIFAC-PapersOnLine, July 2016, pp. 293–299.
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 35 / 36
http://www.bradley.eduhttp://www.jeet.or.kr/LTKPSWeb/uploadfiles/be/201705/290520170957344003750.pdfhttp://www.jeet.or.kr/LTKPSWeb/uploadfiles/be/201705/290520170957344003750.pdf
For Further Reading II
[4] H. B.-P. P. Mndez-Monroy, “Fuzzy control with estimated variablesampling period for non-linear networked control systems: 2-dofhelicopter as case study,” Transactions of the Institute ofMeasurement, vol. no. 7, October 2012.
[5] W. Gao and Z.-P. Jiang, “Data-driven adaptive optimaloutput-feedback control of a 2-dof helicopter,” in American ControlConference, Boston Marriott Copley Place, Boston, MA, USA, July2016.
[6] M. Hernandez-Gonzalez, A. Alanis, and E. Hernandez-Vargas,“Decentralized discrete-time neural control for a quanser 2-dofhelicopter,” Applied Soft Computing, 2012.
G. Janiak, K. Vonckx (Bradley University) 2-DOF Helicopter (Proposal) November 29, 2018 36 / 36
http://www.bradley.edu
IntroductionBackground StudyControl TechniquesModeling a 2-DOF HelicopterPrior Work
Subsystem Level Functional RequirementsBlock Diagram
Preliminary WorkLQR SimulationLQR via USBLQR via WirelessDemonstration
Parts ListSchedule for CompletionFuture Directions
fd@rm@0: