GTMARS

1
GTMARS
Flight Computer Architecture
2
UAV Research LabSchool of Aerospace
EngineeringGeorgia Institute of Technology
Prof. Daniel P Schrage
Suresh Kannan Ilkay Yavrucuk Aaron Kahn
3
Outline of the Video
Introduction to Vehicle Modeling the
Quad-Rotor Flight Computer Control
Systems Navigation Avionics Software Cost and
Reliability
4
Modeling
5
Modeling Procedure
Quad-Rotor Parameters
Nonlinear Model
Aerodynamics Rigid Body Dynamics
Trim Linear Model
Rotor Rigid Body Equations
Open Closed Loop Analysis
Output Rate Feedback Gains, Control Mixing
6
Modeling Hurdles

• No existing models
• No historical data
• Determining effects to be modeled

7
Insert Footage of QR

• Talk about how the scale model helpsed you

8
Model Setup
1
2
Air Vehicle Team Design Parameter Excel Sheet
front
3
4
Simulink
9
Model Effects Included

• Blade Flapping (equivalent hinge offset)
• Quasi-Steady State Tip-Path-Plane Dynamics
• Full 6 DOF Rigid Body Dynamics
• Blade Element Theory
• Arm Dihedral

10
Model Structure
Forces generated by each rotor at its hub
1
2
Resolve Forces And Moments at C.G
Rigid Body Dynamics
3
4
Rotor Arm Lengths
Mass and Inertia Properties
Local Flow velocities seen by each rotor due to
body translational and angular rates
11
Mixing
Nominal omega
Control Mixing
Model
K
Conventional control
rate feedback gain
Input to motor governor
12
Trim
13
Trim, Linearize
Nonlinear Model
14
Longitudinal Hover Matrix
15
Differential Inflow
Pitch Up
Contribution to Pitch Damping, Mq
16
Open Loop Eigenvalues
17
Flight Computer
18
Lander / Earth / Science Payload
Health Sensors
Mission Modes
Vehicle State
Mission Planner
Flight Modes
Position, Velocity,Heading
Trajectory Controller
Kalman Filter Sensor Fusion
Waypoints
Position
Position Control
Attitude Commands
Attitude
Attitude Control
Inertial Sensors
Navigation Sensors
Actuator Commands
Vision / Satellites
19
Mission Planner

• Central Intelligence and Decision Making
• Holds information on global state of World and
  GTMARS internal state
• Plans and Executes Mission depending on its
  Mission Mode
• Configures Hardware and Software for Mission
• Generates Flight Modes during flight

20
Maintain Position
Precision Hover
Do Science Mission 1
Examine Configuration Go back to Lander if
needed Pick up Science Package Configure Software
Configure Hardware/Software
Science Mission 1 Ready
Generate Flight Modes to accomplish Science
Mission 1
Science Mission 1 Mode
21
Mission Modes

• Safe Mode At Lander
• Safe Mode Hover
• Systems Check
• System Startup
• System Shutdown
• Earth Ping
• Take Off
• Precision Hover
• Battery Re-charge
• Science Mission 1
• Science Mission 2

22
error
System Startup
Cycle Comms Hardware
Check Up
Lander Comms
Activate Deployment
Check Rotor Deployment
OK
Cycle Sensors
Check Power Systems
Voltage OK
Check Transponders/Sensors
errors gt n
OK
Safe State
Pre-Flight
Science Configuration
running
error
Motor Rev Up
All Systems Operational
Take-Off Ready Mode
23
Lander / Earth / Science Payload
Health Sensors
Mission Modes
Vehicle State
Mission Planner
Flight Modes
Position, Velocity,Heading
Trajectory Controller
Kalman Filter Sensor Fusion
Waypoints
Position
Position Control
Attitude Commands
Attitude
Attitude Control
Inertial Sensors
Navigation Sensors
Actuator Commands
Vision / Satellites
24
Trajectory Control
Vehicle
Way Points dx, dy, dz
Trajectory Controller
Position and Attitude Control
Flight Modes
Sensors
25
Trajectory Planner

• Takes parameterized flight modes
• Contains a library of flight modes
• Generates a sequence of Way Points for Flight
  Control System

26
Flight Modes

• Position ltx,y,zgt thru ltx,y,zgt
• Come Home
• Precision Hover
• Circumspect ltx,y,zgt radius ltrgt
• Climb ltaltitudegt rate ltaltitude rategt
• Flight Velocity ltVx,Vy,Vzgt

27

• The Mission Planner and Trajectory Controller,
  together, generate low-level way point commands
  for the Flight Control Systems to achieve

28
Control Systems
29
Control Design
Output Quad Rotor Nonlinear Dynamics, Linear
Model
Quad-Rotor Nonlinear Model
Output Stabilized Vehicle Dynamics with
Attitude Control
Inner-Loop Design
Attitude Command System
Output Sensor, Navigation requirements
Outer-Loop Design
Position CommandSystem
Iterate Tweak Attitude Response Requirement to
give good position control
Navigation/Filter Design
Navigation And Filters
Iterate Sensor, Navigation limitations, noise,
variance etc.
30
Control Design Hurdles

• Unmodeled Dynamics
• Uncertainty in the operating environment (Mars)
  affects system dynamics

Adaptive Control
31
Flight Control System
Outer Loop
Inner Loop
?col
?col
?1
?
xcom ycom zcom ?com
?ped
PID/Fuzzy Controller
Model Inversion
?2

Quadrotor Model
PID Controller
?
?lon
M i x i n g
?3
-
?lat
?
?4
Adaptive Neural Network
Attitude and Angular Rates
Position and Velocities
Position Control
Attitude Control (Adaptive)
32
Attitude Command Attitude Hold
Define the pseudo control as
Pitch Angle Command
Kp
Second Order Command Filter
Approximate Inverse Transformation
Quad Rotor Dynamics
Kd
Normalization
Pitch Channel Neural Network
Bias
Sensed States
Pitch Channel Example
33
Neural Network Structure
34
Controller Inputs
Command Filter Parameters
1 2
Filter Damping Filter Frequency
Linear Controller
6 0.8
Proportional Gain Derivative Gain
Neural Network
100 10
Learning Rate Network e-modification (Damping)
35
Pitch Angle Response
36
Network Weights
37
Pitch Rate Response
38
Adaptive Network Output
39
Precision Hover
40
Position Control
Command North 1 meter
41
Pulse Wind Gust of 7 m/shold within 0.3 m
42
Pulse Gust Recovery
43
Constant Wind Gustof 7 m/shold within 0.3 m
44
Constant Wind Gust Recovery
45
Control Systems

• Control authority is adequate for hover and gust
  recovery
• Limits of vehicle not exceeded
• Attitude control is accurate
• Position Control is able to maintain precision
  hover

46
Navigation System
47
Navigation Design
Output System Options
RFP Requirements
Output Pros and Cons
Design Selection
Time of Flight VOR INS Optical
Output Best Design Choice
Analysis of Pros and Cons
Air Vehicle Team Input Propulsion Team Input
Final Design Selection
Simulation and Verification
Iterate Till get most Pros and RFP
requirements are satisfied.
Air Vehicle Team Input Propulsion Team Input
Functionality and Robustness
48
Deployment Scheme
Entry into Atmosphere
Deploy parachute and transponders
Descend vertically
Loose parachute and use rockets to finish landing
Transponders descend
Airbags are retracted and lander deploys
Transponders land and unpack antennae
5 Km
5 Km
49
Transponder
Operation 1st, Wait for Signal Pulse from Lander
or Quad-Rotor. 2nd, Send ID number.
Antenna
I am 5
Solar Panels
50
Navigation Module
To Transponders
Counter 1
Counter 2
Transmitter
Clock
From Transponders
Counter 3
Counter 4
Receiver 1
Decode
Receiver 2
Decode
Receiver 3
Decode
Receiver 4
Decode
51
Transponder Layout
52
Lander Antennae Layout
53
Theory of Navigation
Computations Done in Flight
Initialize the variables
Computations Done on Lander
Compute pseud- ranges
Position of transponder in space by using
antennae on lander and the range measurements.
a,b,c petal 1 antenna pos. d,e,f petal 2
antenna pos. g,h,I petal 3 antenna pos.
Compute variance matrix.
Loop until error is small.
Compute position and error.
54
Navigation Simulator
55
Assumptions

• Lander sitting level
• Petal Length 1m Center 0.2m above surface
• Transponder time delay 0 sec.
• Clock errors are white noise process. Variance
  1e-10 sec.

56
X, Y Error
57
3-D Error
58
Simulated Flight Path
59
Position Error Magnitude
60
Precision Hover
61
Lander-Based Navigation System
Millimeter Wave Radar Range finder
62
Avionics
63
Avionics Design
RFP and Control Requirements
Output Required Data for Control
Output System Devices
Design Selection
IMU Altimeter Nav. Module etc.
Output Final Sensor Suite
Choose off the shelf parts
Air Vehicle Team Input Propulsion Team Input
Check mass and Power req. for Sensors
Mount Devices on GTMARS
Air Vehicle Team Input Propulsion Team Input
Sensor Integration
64
Main Power Distribution Bus (multiple voltages)
Motor Sensors
Health Sensors
Peripherals
IMU
Computer
Alt.
Science Payload
Navigation Module
Transponder Transceiver
Compass
Telemetry Systems
Clock
65
Main Power Distribution Bus (multiple voltages)
Used to provide a link between the fuel cell and
the other devices on the GTMARS. Mass (all
converters) 0.3 Kg Power (output) 20.1 watts
66
Science Payload
Used to gather information from the environment
not needed to fly the vehicle. Consists of 2
parts, fixed cameras on GTMARS, and removable
modules carried aboard. Mass (both) 2.3 Kg
Power 10 watts
Science Payload
67
Health Sensors Group
Health Sensors
Used to monitor temperature, voltage,
current, stress, and fuel cell properties to make
sure that no values exceed nominal operating
limits. Mass (all health sensors) 0.04 Kg
Power 0.1 watts
68
Computer and Peripheral Devices
Peripherals
Used to communicate and gather data from other
devices on GTMARS. Mass (both) 0.640 Kg
Power 10 watts
Computer
Used to send and receive data from the lander.
Built in 2 parts, short range high bandwidth,
and long range low bandwidth. Mass (both)
0.362 Kg Power 2 watts
Telemetry Systems
69
Navigation Module
Used to measure position of vehicle in X, Y
direction. Mass 0.521 Kg Power 2 watts
Navigation Module
Transponder Transceiver
Clock
70
Stability and Control Sensors
Motor Sensors
Used to regulate commanded rotor RPM Mass
0.001Kg Power 0.005 watts
IMU
Used with Navigation system to measure vehicle
attitude Mass (IMUCompass) 0.91 Kg Power 3
watts
Used to measure vehicle altitude above the
ground Mass 0.453 Kg Power 3 watts
Alt.
Compass
Used with IMU to measure vehicle heading (see IMU
for joint information)
71
Current Avionics at Georgia Tech
72
Flight Computer
73
Telemetry/Sensors
74
Test Aircraft
75
Software Architecture
76
Software Design
Output Generated Code Module
Algorithm Development
Output Generate Wire up Data File
Algorithm Wire-up
Hookup modules QoS etc
Domain Specific Tool eg. Simulink
Output Production Code
OCPNDDS
Hardware In Loop
Implement on Flight Computer for Testing
Fix Bugs Apply JPL Checking Procedures Units
Check !
Production Code
ControlShell
77
Software

• Modular Algorithm Objects
• Quality of Service
• Autocoding
• Late wire-up of Components
• Reconfigurability
• Team Oriented Integration Software
• Version Control

78
Algorithm Object
Input Ports
Output Ports
Input Port Description DataType position QoS
maxrate 10Hz minrate 1Hz criticality
HIGH bandwidth HIGH
Algorithm Object Generated from domain Specific
tool
79
Software Architecture
World Sensing
Science Data
Algorithms
Mission Modes
Mission Satisfaction
Open Control Platform
Quality of Service
Scheduled Messaging
VxWorks
Science Algorithms
Real-Time
Network Hardware
Messaging position_at_uavcommand_at_innerloopflightmo
de_at_uav
Quality of Service maxrate 10Hzminrate
1Hz criticality HIGH bandwidth COLLOCATED
Algorithms Navigation, FiltersFlight Control
SystemTrajectory Controller Mission
PlannerScience Algorithms
80
Repository
Algorithm Components
Wire-Up
81
Software at Georgia Tech

• Used on Unmanned Helicopters
• Flight Controls Research
• Quick Development Time
• Highly Networked

82
Insert Footage of Heli
83
Cost and Reliability
84
Cost and Reliability

• R and D time given from Air Vehicle Design team
  requirements - 6 man-years
• R and D budget given from Air Vehicle Design team
  requirements - 1.2 M
• Military Standards that electronics must satisfy
  for system reliability
• Altitude MIL-STD-810C, Class 2, Method 504.1
• Vibration MIL-STD-81-C, Method 514.2, Category
  (e), Procedure V, Part 1, Curve Q, Part 2, Curve
  AF
• Shock MIL-STD-810, Method 516.2, Figure 516.2-2,
  Procedure I, Amplitude (a), time (c),
  Non-operating
• Salt Atmosphere MIL-STD-810C, Method 509.1,
  Procedure I
• Acceleration MIL-STD-810C, Method 513.2,
  Procedure II, Operating 1 minute at 7.5g, 6
  directions.
• Temperature Operating -54oC to 71oC
  Non-operating -57oC to 85oC

85
Cost Breakdown

• Initial Integration and System development
• 3 man-years
• Final Integration and Software development
• 1.5 man-years
• Flight Tests of Prototype Vehicles
• 1.5 man-years

86
Cost and Reliability

• R and D time given from Air Vehicle Design team
  requirements - 6 man-years
• R and D budget given from Air Vehicle Design team
  requirements - 1.2 M
• Military Standards that electronics must satisfy
  for system reliability
• Altitude MIL-STD-810C, Class 2, Method 504.1
• Vibration MIL-STD-81-C, Method 514.2, Category
  (e), Procedure V, Part 1, Curve Q, Part 2, Curve
  AF
• Shock MIL-STD-810, Method 516.2, Figure 516.2-2,
  Procedure I, Amplitude (a), time (c),
  Non-operating
• Salt Atmosphere MIL-STD-810C, Method 509.1,
  Procedure I
• Acceleration MIL-STD-810C, Method 513.2,
  Procedure II, Operating 1 minute at 7.5g, 6
  directions.
• Temperature Operating -54oC to 71oC
  Non-operating -57oC to 85oC

87
Final Slide

• Vehicle Concept is Feasible
• Hierarchical Control
• Controls Response Rates Adequate for Maneuvering
• No Technology Extrapolations Needed
• Open Software Architecture