May 28th, 2008, Stuttgart
Georeferencing with
AEROcontrol GPS/IMU data
Jens Kremer
IGI mbH
57223 Kreuztal / Germany
IGI mbH
Langenauer Str. 46
57223 Kreuztal, Germany
www.igi-systems.com
Guidance and
Sensor Management
GPS / IMU Systems
Integrated Sensor Systems
for Special Applications
Guidance and
Sensor Management
GPS / IMU Systems
Integrated Sensor Systems
for Special Applications
CCNS:
Guidance and Sensor Management
Development in Aerial Survey
Today
1988
Guidance and
Sensor Management
Integrated Sensor Systems
for Special Applications
GPS / IMU Systems
• LiteMapper
• DigiCAM
• StreetMapper
Airborne LiDAR Mapping
digital elevation
model (DEM)
point cloud
LiteMapper - Airborne Lidar Mapping
pulse repetition rate: up to 240 kHz
range: up to 1800 m
number of targets / pulse: unlimited for digitized waveform
Installation in a Bell 206 Jet Ranger
Installation in a MI 8
380KV Powerline h = 85m / v= 40kn
380KV Powerline h = 85m / v= 40kn
Project “Aahlen”
Flood Risk Management
Simulation of Elbe river water levels
Project: city of Dresden, Saxony, Germany
DSM Digital Surface Model
Ground point spacing:
1 point/m² raster, approx.
4 mio. points total
Images kindly provided by Milan Flug GmbH
Image © Landes tals perrenverwaltung (LTV),
Fre ista at S achsen, Germany
Flood Risk Management
114 m
Flood Risk Management
116 m
Flood Risk Management
118 m
Flood Risk Management
120 m
DigiCAM
Corridor mapping
Power lines
Pipelines
Roads
Rail tracks
Small to mid size projects
Rapid response applications
DigiCAM-H/39
39 Mpixel CCD back
Control computer with two storage units
Graphical User Interface with touch screen
DigiControl GUI
DigiCAM Installation Examples
DigiCAM Installation Examples
Dual DigiCAM
Dual DigiCAM
Total image width (across track):
• One exposure = two separate photos
~ 14 kPixel or
~ 10,4 kPixel
Untersuchung der Genauigkeit II
Untersuchung der Genauigkeit II
Guidance and
Sensor Management
GPS / IMU Systems
Motivation
Integrated Sensor Systems
for Special Applications
- Which accuracies do we need?
Why GPS and IMU?
GPS/IMU integration
The AEROcontrol system - used components
How to use GPS/IMU results for Georeferencing
Examples
Direct Georeferencing
connect information of
an airborne sensor with a
position in space
Georeferencing
Indirect using the sensor data
Example: AT
Z
Direct
Y
NOT using the
sensor data
position
attitude
velocity ( e.g. SAR)
X
Why GPS/IMU?
Accuracy : What do we need?
Position Attitude
As good as the
aspired accuracy of the
final result
~ cm to m
Good enough to reach
cm to m accuracy on
the ground:
e.g. TOP10DK
image scale: 1:25000 (3825m)
α
h
tan (α)= dx / h
tan (α)= 1m / 3825m
α= 0.015°
dx
~ 0.05° to 0.001°
dGPS
GPS: ~5-10m
Differential GPS using phase and doppler measurements:
Possible accuracy :
a few cm (kinematic!)
Satellite
Dependent on:
Distance from the base station
Satellite configuration (number
and position of the satellites)
Reception conditions
Rate 1 - 20 Hz
Reference
Station
No attitude measurement
(antenna array:
5m distance, 1 cm ∆pos:
∆α > 0.1°)
Inertial Navigation
High data rates: > 50 Hz
Attitude measurement as good as you like ( ... price?! )
“Dead reckoning system”
Positioning Accuracy [m]
All sensors are imperfect => increasing attitude and position error:
e.g. accelerometer bias ab => position error = 1/2 ab t2
Time [min]
GPS/IMU Integration
1. Calculate initial position, attitude and velocity (the initial state)
2. Calculate the actual state with the IMU data (“strapdown algorithm”)
“Strapdown Algorithm”
initial state
gyro
attitude
earth rotation and
rotation of the navigation
coordinate system
position
acc.
gravity
coriolis
correction
e.g. 256 Hz
velocity
GPS/IMU Integration
1. Calculate initial position, attitude and velocity (the initial state)
2. Calculate the actual state with the strapdown algorithm
3. When GPS measurements occur: use Kalman Filter to
estimate optimal state, including IMU properties
state vector
x: pos, att, vel, IMU properties
measurement vector
y: ∆pos, ∆vel
GPS/IMU Integration
1. Calculate Initial Position, Attitude and Velocity (the initial state)
2. Calculate the actual state with the strapdown algorithm
3. When GPS measurements occur: use Kalman Filter to
estimate optimal state, including IMU properties
state vector
x: pos, att, vel, IMU properties
measurement vector
y: ∆pos, ∆vel
Note:
forward- backward calculation and smoothing
can improve the result significantly!
AEROcontrol
Differential GPS
Inertial Measurement Unit
Direct
Inertial Aiding
Kalman Filter
Position Velocity Attitude
AEROcontrol-IId
Hardware
AEROcontrol computer unit with
GPS receiver and IMU
Software
AEROoffice software incl. GrafNav
AEROcontrol-IId IMU
• Three accelerometers
• Three gyroscopes
Measurement of the motion in all three axes.
IMU - Inertial Measurement Unit
IMU with fiber-optic gyros
Fiber Optic Gyros
The Fiber Optic Gyro Principle
1.
1. no
no Rota
Rotation
tion
2.
2. with
with Rotation
Rotation
© LITEF GmbH 20 05
IGI IMU-IId
500m optical fiber
AEROcontrol IMU - IId
with fiber optic gyros
Drift: 0.1°/h
Noise: 0.02°/ SQRT(h)
Rate: 128 or 256 Hz
GPS Receiver
NovAtel OEMV-3
•
•
•
•
72 channels
triple frequency
optional OMNISTAR HP
optional GLONASS
Precise Positioning
Post processed differential GPS:
• own station
• public station (“CORS”)
Real time differential GPS:
PPP - Precise Point Processing:
• StarFire - NavCom SF-2050M
•
OmniStar HP
integrated in AEROoffice with
GrafNav 8.10
dGPS: with Base Station
Post Processing
- use your own station or
- data from permanent
stations (CORS):
www.sapos.de
OmniSTAR HP
OmniSTAR HP coverage
AEROcontrol GPS board incl.
NovAtel GPS with OmniSTAR-HP
OmniSTAR HP
position and orientation
available before landing
North
East
Up
Mean
RMS
-0.231 m 0.034 m
-0.256 m 0.051 m
-0.174 m 0.069 m
PPP - Precise Point Processing
integrated in AEROoffice with
GrafNav 8.10
Lever Arm Correction
1.75m
Lever Arm Correction
Lever Arm Correction
wind
tan 5° * 1.75m = 15cm
Lever Arm Correction
• Mount angles read by AEROcontrol
• Changing lever arm calculated during post processing
Lever Arm Correction
Plots: IMU position prediction vs. GPS position measurement
with mount
angle readout
without mount
angle readout
Direct Georeferencing
Georeferencing
connect information of
an airborne sensor with a
position in space
Indirect using the sensor data
Example: AT
Direct
NOT using the
sensor data
Z
position
attitude
velocity ( e.g. SAR)
Y
at the correct time!
correct sensor model
X
System Calibration
GPS Antenna
Exact synchronization
The lever-arms have to
be taken into account
zB
Careful measurement
of the attitude of the
IMU
xB
zS
Sensor
yS yB
IMU
xS
Correct sensor model at
the data collection time
Calibration
System Calibration
Calibration
IMU Calibration
Sensor Calibration
Misalignment Calib.
• IMU properties
• focal length
• principal point
• angle encoder
• ...
• angle offset
• position offset
• at the IMU provider
• IGI: 2 years
• Calibration lab
• camera: 2 years
• calibration field or
mission area
• after system
modification
• or selfcalibration
DG or ISO ?
Position and Attitude
from GPS/IMU
DG
Georeferenced
Image Data
ISO
Direct Georeferencing
Integrated Sensor Orientation
Direct use of calibrated
GPS/IMU data to georeference
image data.
Extended aerial triangulation with
GPS/IMU data as additional input.
Direct Georeferencing
No GCPs and no AT
Very fast
Need for calibration
Z
No redundancy
Y
X
Limited accuracy
Integrated Sensor Orientation
Need for AT of the
mission area
No extra calibration
Z
Redundant measurement
Accuracy not limited by
GPS or IMU
Y
X
Integrated Sensor Orientation
• extended AT for the mission area
• GPS/IMU results are used
as additional measurements
• without any ground control
• with 1 GCP
• with some GCPs
• no need to determine calibration
parameters explicitly
Direct Georeferencing I
• extended AT for a small part
of the mission area
• without any ground control
• with 1 GCP
• with some GCPs
Direct Georeferencing II
• use calibration from an other area
(e.g. flown at an other day)
DG or ISO ?
Displacement of a pixel from position and attitude errors:
dx
pitch
yaw
Roll / Pitch
roll
dSposition
dSangle
Example:
DMC with 12µm pixels and 120mm focal length
tan 0.004° * 120mm ≈ 8µm
1RMS angle error < 1 Pixel
DG or ISO ?
Displacement of a pixel from position and attitude errors:
dx
pitch
yaw
Roll / Pitch
roll
dSposition
dSangle
Example:
Aerial photo mission for large scale maps
flying height ~500m
tan 0.004° * 500m = 3.5cm
dGPS accu racy ??
DG or ISO ?
Position and Attitude
from GPS/IMU
DG
Georeferenced
Image Data
ISO
Direct Georeferencing
Integrated Sensor Orientation
Direct use of calibrated
GPS/IMU data to georeference
image data.
Extended aerial triangulation with
GPS/IMU data as additional input.
small to medium scale mapping,
orthophotos
medium to large scale mapping
• largest savings possible
• highest accuracy
• no extra boresight calibration
The optimal choice of the georeferencing workflow
(and calibration) is the key to a (commercially)
successful operation of an aerial camera/GPS/IMU
system.
Mission Examples
DG:
Dual-DigiCAM ifp / IGI
ISO: strip project, RWS highway
DG:
UltraCam block project "Wesco"
Dual DigiCAM
Example: Direct Georeferencing
• optimal GPS conditions (GPS accuracy ≤ 5 cm)
• good lever arm determination
Estimation of the angle- and position accuracy:
0.004 °
dSangle ≈ tan 0.004° • 850m = 6 cm
hg=850m
dSposition ≈ 5 cm
dSangle
071219 - Vaihingen Enz
19.12.2007
Weser Bildmessflug
Cessna 206
(single engine)
Weather Conditions
Mission Planning 7cm GSD
7 cm GSD
60% / 76%
Direct Georeferencing Results
Preliminary results!
Final results will be presented at the XXI ISPRS Congress 2008 in Beijing.
Integrated Sensor Orientation
Aerial photography of a track
AT:
The roll angle is difficult to get,
the yaw angle is well known.
Consequence:
additional parallel lines
a high number of GCP‘s
GPS/IMU: The roll angle is very well known, the yaw angle is more
difficult to get.
For a single strip, the AT and GPS/IMU complement one
another very well.
Integrated Sensor Orientation
Integrated Sensor Orientation
ISO with 88 GCPs
SIGMA 0 =
4.07 (1/1000)
RMS control point residuals: 18.
Max. control point residuals: 49.
6.
17.
7. (1/1000)
27. (1/1000)
Integrated Sensor Orientation
ISO with 4 GCPs
SIGMA 0 =
3.94 (1/1000)
RMS control point residuals: 12.
Max. control point residuals: 15.
2.
2.
RMS of check point
residuals:
32.
38.
50.
7.
12.
Integrated Sensor Orientation
0,2
0,15
X [east]
RMS: 1.5 cm
0,1
4
8
5
4
6
3
4
4
1
4
1
9
3
9
7
3
7
5
3
5
3
3
3
1
3
0
9
2
8
7
2
6
5
2
4
3
2
2
1
1
9
9
1
7
7
1
5
5
1
3
3
8
9
1
1
1
6
7
4
5
1
0
2
3
0,05
-0,05
-0,1
-0,15
-0,2
0, 2
0, 15
0, 1
495
476
457
438
419
400
381
362
343
324
305
286
267
248
229
210
191
172
153
134
115
9
6
7
7
5
8
3
9
1
0
2
0
0, 05
Y [north]
RMS: 2.7 cm
-0, 05
-0, 1
-0, 15
-0, 2
0, 2
0, 15
0, 1
-0, 05
-0, 1
-0, 15
-0, 2
50
1
48
1
46
1
44
1
42
1
40
1
38
1
36
1
34
1
32
1
30
1
28
1
26
1
24
1
22
1
20
1
18
1
16
1
14
1
12
1
10
1
8
1
6
1
4
1
1
0
2
1
0, 05
Z [height]
RMS: 3.0 cm
Aerial photography mission flown for
Date:
September 21st 2006
Area:
Tsuruma (near Tokyo)
Camera:
ULTRACAM D
Camera mount
leveling:
AEROcontrol
Guidance and
precise positioning:
CCNS/AEROcontrol
Data processing:
MATCH-AT and AEROoffice
WESCO: Mission Area
Greater Tokyo Area
Airport of departure:
Chofu Airport
17 sqkm
WESCO: IMU Installation
WESCO: Camera and GPS Antenna
WESCO: Flight Planning
GSD:
15cm
Block:
4*16 Images
Overlap: 80% / 60%
13 Ground Control Points
GSD:
7cm
Block:
4*28 Images
Overlap: 80% / 60%
WESCO: Flight Mission
September 21st 2006
WESCO: Guidance Accuracy I
2m
33m
0.07m
0.01m
35m
35m
WESCO: Guidance Accuracy II
IFD
AFD
2.31
17.66
Mean –0.43
2.64
RMS
Difference [m] in flight direction (caused by the CCNS and the real-time GPS accuracy).
Difference [m] perpendicular to the flight direction (caused by the pilots skills).
WESCO: Misalignment Calibration I
WESCO: Misalignment Calibration II
Automatic tiepoint generation with INPHO MATCH-AT
WESCO: Misalignment Calibration III
Misalignment calibration inside AEROoffice
∆ωφκ < 0.004°
WESCO: Direct Georeferencing
10003
10005
10008
INPHO
MATCH-AT 5.0
©
M
nG
li a
rv
se
eo
ic
e
www.igi-systems.com