Adaptive Flight Control of a Sensor

Guided MK-82 JDAM

Kevin A. Wise, Ph.D.

Senior Technical Fellow
Integrated Defense Systems
SAE Mtg, 12 October 2006, Williamsburg VA
 Outline

• Joint Direct Attack Munition (JDAM)

• Laser-JDAM MK-82
• Adaptive Control Overview
• Flight Test Results
• Movies
• Open Problems in Adaptive Control
of Aircraft and Weapons
• Summary

2
 JDAM Tail-kit

• JDAM (Joint Direct Attack Munition) is a tail-kit for

“dumb” bombs that provides:
– Actuated fins
– Guidance and control software
– GPS/INS navigation
– Strakes to improve aero

3
 The JDAM Weapon Family
Affordable, Accurate, Autonomous, Adverse Weather

MK-84
(2,000 lb)
BLU-109
(2,000 lb) MK-83, BLU-110
(1,000 lb) MK-82, BLU-111
*Currently in Developmental Test and Evaluation (500 lb)*

4
 Baseline JDAM Free-Flight
Timeline
Altitude

Optimal Guidance Phase

Impact Phase
Transfer Alignment

Separation Phase

Roll Over to
Pull Down
on Target

Time

T=0 T=1 sec T=3 sec T=22-30 sec T=1 sec to go

•Release •Unlock •Start GPS (24 sec typ) •Drive AOA Target Impact
•Start Fins Search •First to Zero
Guidance •Start Navigation
Autopilot Update

6
 Guidance Law Control of Impact Angle

Impact Angle Commands: 30, 40, 50, 60, 70

Release
20000

15000
Altitude (ft)

10000

5000

0 Target
0 10000 20000 30000 40000
Down Range (ft)

7
 Baseline Control:
Feedback Gains Designed Using
Optimal Control + Projection Theory

• AUTOGAIN Tunes LQR

Optimal Robust Servomechanism Linear Parameters
Quadratic Regulator (RSLQR) • Convergence Criteria Focus
~ ~ On Stability/Actuator Rates
Model: z• = Az + B zRn • LQR Design Charts Describe
~ ~ ~ ~
ARE: ATP+PA+Q-PBR-1BTP = 0 Tuning Process

CLAW: u=-KSFx= -R-1BTPx

Optimal Projection To Output
Augment Dynamics With Feedback Architecture
Integral Control For Perfect • Select Dominant
Command Tracking S-Plane
Eigenstructure (r,Xr), r<n
X X • Project Gains (Static)
• Preserve Excellent Stability XX K=KSFXr(CXr)-1
Properties Of State Feedback 0 XX 0 u=-Ky
XX • Analyze Output Feedback
Using Output Feedback
X X Design
• Eliminates Sensor H/W • Iterate LQR To Achieve
Required For State Feedback Desired Bandwidth

8
 JDAM Greatest Hits Vol 1

Surface
Target

Buried Target

Delay Fuze for

Underground
Detonation

Approved for Public Release 9 Oct 1998 9

JDAM Greatest Hits Vol 2

10
 AOA Collapsed to Zero at Impact

MK-84 JDAM Just Before Impact

Hole Shows Fins,

Strakes, Strap
Tensioning Screws,
Launch Lugs

Approved for Public Release 9 Oct 1998 11

 Laser JDAM Program
• Laser JDAM adds a laser seeker to the baseline
MK-82
– Laser designator is used to paint target
– Weapon flies optimal GPS/INS to fixed coordinates until
laser sensor is in range
– After laser acquisition, weapon guides to target
• Added seeker hardware + raceway for wire
harness cause Laser MK82 aerodynamics to differ
from the baseline
• Adaptive control augmented to the baseline MK-
82 autopilot for the MK82 Laser to compensate for
the differences

12
 Adaptive Control Transitioned To
Advanced Weapon Systems
• Adaptive Control Based upon Earlier Aircraft Application
Adaptive Flight Control
– Extended to Munitions (00-02) with GST
– Boeing IRAD Improvements Focus on System ID,
Implementation, and Actuator Saturation Issues
– Design Retrofits Onto Existing Flight Control Laws
Boeing Collaborates
– Flight Proven on MK-82 L-JDAM, (04-06) AFOSR Adaptive With Prof. N.
Control of UCAVs I,II Hovakimyan at
– Transitioned To Production JDAM VaTech on limited
actuation
Technology Transition Timeline
93 94 95 96 97 98 99 00 01 02 03 04 05 06
Boeing funds MIT
(Dr. A. Annaswamy)
Intelligent Flight Adaptive Control For Boeing IRAD to initiate research in

Control System Munitions V&V of adaptive

systems
(NASA/Boeing) (AFRL-MN/GST//Boeing)
F-15 ACTIVE MK-84 MK-82 L-JDAM

Reconfigurable
Control For Tailless
• Gen I, flown 1999, 2003
Fighters MK-84 JDAM
• Gen II, 2002 – 2006 (AFRL-VA/Boeing)
•flight test 4th Q 2005 X-36
• Gen III, 2006
• Ongoing NASA/Boeing IFCS
• Other Transitions
13
 Adaptive Augmentation
• Retrofits onto an existing autopilot (baseline A/P unchanged)
• Baseline A/P commands incremented/decremented as needed
• Uses a reference model representing the desired closed-loop
dynamics
• Adaptive increment makes airframe behave like the reference
model
• Adaptation dormant while airframe response matches reference
model to within pre-specified tolerance
• Provides robustness to
modeling errors (aero
Reference +
Model
-
uncertainties) Adaptive
Control

Optimal Baseline + +
Airframe
Guidance Autopilot

14
 MK-82/L Adaptive Autopilot
• Baseline JDAM autopilot
– LQR PI with output projection
– High confidence design, tested extensively and in production
– Constructed using wind-tunnel data and gain-scheduling

• Adaptive augmentation
– Developed for the Laser JDAM demonstration program
– Allowed baseline MK-82 autopilot (and gains) to be applied to MK-82
Laser
– Later added to the MK-82 baseline autopilot
– Autopilots of both MK-82 variants now use the same autopilot
architecture and gains (including the adaptive components)
– Direct-adaptive control
– No off-line training

15
 Generalized Plant and Baseline Controller
Open-Loop Dynamics



 x p  Ap x p  B p  u  f  x p    x p   Ap

   B FC
0   x p   Bp 
      u  f 
x 
 0 
 p B r
x A x B F D
c   2 c   p  c  c   2 c   p 
 x c  Ac xc  B1c r t   B2 c F y 1c 
 x A x B1 B2
 y C p x p  D p  u y C p 0  x  D p u C x  D u
    
u K x p x p  K xc xc  K r r t 
T T T D
C

system state controller state

moment uncertainties
extended system state
f  x  T   x    f  x 
control failures


x  A x  B1  u  f  x p   B2 r t  
inner-loop commands guidance commands

16
 Reference Model and Adaptive Control

• Set uncertainties to zero:  0mm , f  x p  0m1 Reflects

Desired
• Use baseline A/P: u ubl K xT x  K rT r t  Weapon
Dynamics

• Formulate Closed – Loop System Dynamics

x ref  A  B1 K xT  xref  B2  B1 K rT  r t   Aref xref  Bref r t 
         
Aref Bref

– defines nominal closed-loop dynamics achievable under baseline A/P

– forms desired dynamics for adaptive augmentation with uncertainties
ˆ T  x 
u ubl  uad K xT x  K rT r t   kˆxT x  kˆrT r t   
              
p

• Control: Baseline A/P Adaptive Augmentation

    r t   ˆ  x 
T T
u  K x  kˆx x  K r  kˆr T
p

17
 Parameter Adaptation
• Theoretical Basis
– 2nd Theorem of Lyapunov
– Barbalat Lemma
– Universal Approximation Property of RBF NN
• Adaptive laws yield bounded tracking performance with all signals
bounded, in the presence of uncertainties, (UUB) Control Weapon
kˆ  Proj
 x x kˆ ,  x e P B 
x
T
1
Response Through
Reference Model.
 ˆ
kr r Proj kˆ ,  r t e P B 
z
T
1
lim x  xref C
t 
 Uniform Response
ˆ


 Projˆ ,  x e P B 
p
T
1 For Each Weapon

x t   xref robustness
– using Dead-zone modification, (enforces t   to noise)
 freezes adaptation if:
– using Projection Operator, (bounds adaptive parameters)
– using e – modification, (adds damping and bounds adaptive parameters)
– using μ– modification, (protects against control saturation)
18
 Adaptive Augmentation of RSLQR
Optimal Pitch Autopilot

Incremental Elevator

a z azcmd q Adaptive Control  ead

Command

Turn Rate qcmd

a  r Vehicle
Inner Loop
AZcmd + + + e 3
KAZ KI 1/s Fin Mixing Actuators
- - -
AZ KP 600 Hz
Cperc q 3rd Order
Elliptic Filter

Lever Arm * s IMU

-
1st Order Lag Mean AZ 3rd Order
Noise Filter + Filter Elliptic Filter

100 Hz
19
 Adaptive Augmentation of
RSLQR Optimal Roll-Yaw Autopilot
Incremental Ail/Rud
Commands
 err ay a cmd
y ps rs  Adaptive Control  ad
a  rad 

Turn Rate rcmd

Inner Loop
Error + + a Vehicle
KPHI 1/s
- 3
- KI Fin Mixing Actuators
Aycmd=0 + + lead-lag r
KAY 1/s filter
+ - -
- e
AY KP
Cperc ps rs 600 Hz
p 4th Order
Transform to Elliptic Filter
Stability Axes
r 4th Order
Elliptic Filter IMU
Lever Arm * s

1st Order Lag

- Mean AY 4th Order
Noise Filter + Filter Elliptic Filter
100 Hz 20
 Simulation-Based Evaluation

• Trajectories with both open and closed-loop guidance

• Monte-Carlo Testing:
– Aerodynamic uncertainties
 Body forces and moments and fin moments (no fin forces)
– c.g. location uncertainty in all three axes
– Winds and turbulence
Histogram
Histogramofofmean
40
meanaerodynamic
aerodynamicperturbation
perturbation

• Results show adaptive a/p 35

provides added robustness 30

Number of Cases
Number of Cases
25

20

15

10

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Mean Aero Perturbation

Mean Aero Perturbation

21
 Robustness to Time Delays
• Time delay sensitivity evaluated via simulations (nominal aero)
– Sweep through various combinations of input and output time-delay
– Simulation time-histories “eyeballed” to determine goodness and given
values based on amount of activity
• Results show more than adequate time-delay margins
Adaptive A/P Time-Delay Sensitivity Nominal A/P Time-Delay Sensitivity
40 40
Adaptive a/p Nominal a/p
35 35

30 30
Input Delay (msec)

Input Delay (msec)

25 25

20 20

15 15
Nominal hardware
time delays below 10 10
minimum of chart
5 5
5 10 15 20 25 5 10 15 20 25
Output Delay (msec) Output Delay (msec)

22
 Oct 04 Flight Data (1 of 2)
Bank Maneuvers
+30 deg -30 deg

AOA

Beta

Qbar

Mach

+BETA -BETA 23
 Oct 04 Flight Data (2 of 2)
Roll Maneuver
+30 deg -30 deg
Dele (deg)
Dela (deg)
Delr (deg)

+BETA -BETA 24
LJDAM - Jan 05 Fixed Target

25
LJDAM – May 05 Moving Target

26
LJDAM – May 05 Moving Target
Remote Controlled Target

27
MK82 Laser SDD-G2

28
MK82 Laser – 40 mph HMV

29
 Lessons Learned

• X-36 RESTORE Flight Test

X-36 – Stabilized Unstable Airframe Under Significant Failures
– Limited Flight Envelope

• MK-84 JDAM Dynamic Inversion CLAW

– Eliminated Gain Scheduling Requirements
JDAM
– Used Existing Truth Model for Analysis/Comparison

• MK-82 LJDAM Augmented LQR

– Retrofit Onto Baseline Control
– Significant Parameter Tuning Required For Performance

Flight Results Have Created List of Open Problems

30
 Open Problems

• Reference Model Design

• Parameter Tuning Guidelines
• Adaptive Dead-zone and Learning Rates
• Adaptive Structural Mode Suppression
• Gain and Phase Margins for Adaptive Systems
• Retrofit For Legacy Systems

31
 Summary

• DOD Requires Robust System Behavior

for Autonomous UAV and Weapon
System Operation – Need for Adaptive
Control
• Flight Quality Computer Hardware Now
Capable of Advanced Algorithms
• Industry Actively Maturing Technology

32