Generatinggy LiDAR data in laboratory: LiDAR...
Transcript of Generatinggy LiDAR data in laboratory: LiDAR...
Generating LiDAR data in laboratory: g yLiDAR Simulator
Bharat Lohani, R K MishraDepartment of Civil EngineeringDepartment of Civil EngineeringIndian Institute of Technology Kanpur Kanpur INDIA
M i iMotivation…
Are LiDAR data available ?
LiDAR data NOT available in majority ofLiDAR data NOT available in majority of countries
Lack of awarenessLack of awarenessSecurity issuesCost
Bharat Lohani, IIT Kanpur India
Are LiDAR data available ?
LiDAR data NOT available for teachingLiDAR data NOT available for teaching purposes
Readily available dataReadily available data Data with as-desired specificationsData with ground truthg
Bharat Lohani, IIT Kanpur India
Are LiDAR data available ?
LiDAR data NOT available for researchLiDAR data NOT available for researchData with a wide range of desired specificationsData with complete and 100% accurate groundData with complete and 100% accurate ground truth
Bharat Lohani, IIT Kanpur India
Solution lies in LiDAR simulator…
User creates a terrainUser creates a terrainUser chooses the flight parametersLiDAR data are generated for createdLiDAR data are generated for created terrain as if the actual LiDAR sensor had flown the terrainflown the terrain
Bharat Lohani, IIT Kanpur India
Design consideration forDesign consideration for simulator
Should be . . .
User friendlyUser friendly
Wider distribution
Help or tutorial
Bharat Lohani, IIT Kanpur India
Can simulate . . .
Generic sensor
Specific sensorsSpecific sensorsALTMALSAnd others…
Bharat Lohani, IIT Kanpur India
Should simulate trajectory as in a normal flight
6 degrees of freedom6 degrees of freedom
Bharat Lohani, IIT Kanpur India
Should simulate earthlike surfaces
Source: Optech Inc.Bharat Lohani, IIT Kanpur India
Also…
Possibility of error introductionPossibility of error introduction
Output data available in common formats
Bharat Lohani, IIT Kanpur India
Development of simulatorDevelopment of simulator
System components
IntegrationSensor component
T i
Trajectory
Terrain component Out
putInputTrajectory component
Bharat Lohani, IIT Kanpur India
T j t p tTrajectory component
Flight directionLocation Attitude
Location
Location: coordinates of laser head at each firing of pulse
Location depends on Instantaneous accelerationsaccelerations
Instantaneous accelerations should be simulated as in a normal flight: pseudo-random simulation
Bharat Lohani, IIT Kanpur India
Acceleration simulation
2 2sin ( ( )) cos ( ( )) ( )J K
iX j j t k k t ta A B id C D id m id
T Tπ π⎛ ⎞ ⎛ ⎞= + +⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠∑ ∑
1 1X j j t k k t t
j kT T= =⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠
∑ ∑
F = Firing frequencyJ,K,A,B,C,D and m governing parametersparameters
Bharat Lohani, IIT Kanpur India
Location simulation1 21
2i i i i
x t x tX X u d a d+ = + +2
X
V l it i di ti fli ht i X iux = Velocity in direction flight i.e. X axis
Bharat Lohani, IIT Kanpur India
Attitude (Roll, Pitch, Yaw) simulation
2 2sin ( ( )) cos ( ( )) ( )J K
ij j t k k t tR A B id C D id m id
T Tπ π⎛ ⎞ ⎛ ⎞= + +⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠∑ ∑
1 1j j t k k t t
j kT T= =⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠
∑ ∑
Bharat Lohani, IIT Kanpur India
Sensor componentsSensor components
Sinusoidal scan patternpZig-zag scan pattern
Sinusoidal scan patternpLet time taken to complete 1/4th of a scan is T.
P is the numbers of points in 1/4th of a scan.
( )
p
The maximum scan angle is өmax.
1
( )max 2sin
where
i iT
Ti i
t
t
πθ θ=
0
0.5
Z (m
)
where, i iPt =
160140160
-1
-0.5
6080
100120
140160
80100
120140
X (m)Y (m)Bharat Lohani, IIT Kanpur India
Zig-zag scan patterng g p
θ max iPiθθ =
1
0
0.5
Z (m
)
140160
-1
-0.5
6080
100120
140
80100
120140
X (m)Y (m)Bharat Lohani, IIT Kanpur India
Terrain componentTerrain component
Modeling surfaces: earthlikeVector approach: mathematical surfacesRaster approach with over ground objectsFractal terrain
Vector approach: Mathematical surfacespp
Simple SurfaceSimple Surface•AX+BY+CZ+D=0
Example of a simple surface: 2X+5Y+10Z-100=0 (displayed in surfer)Bharat Lohani, IIT Kanpur India
Vector approach: Mathematical surfacespp
Complex Surfacep•Z=A[sin(X/B)+sin(XY/BC)]+D
Example of a complex surface: z=10[sin(X/25)+sin(XY/(25*50)]-300 (displayed in surfer)Bharat Lohani, IIT Kanpur India
Raster surfaces
Bharat Lohani, IIT Kanpur India
Fractal surfaces
Bharat Lohani, IIT Kanpur India
Integration of componentsIntegration of components
Integration of components
i i iX X Y Y Z Zi i i
i i i
X X Y Y Z Za b c− − −
= =
Bharat Lohani, IIT Kanpur India
Error introduction in simulated data
2( )i iX X N μ σ= + ( , )T t X XX X N μ σ= +
Bharat Lohani, IIT Kanpur India
Concept implementationConcept implementation
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Optimal flight lines given by systemp g g y y
Bharat Lohani, IIT Kanpur India
User defined flight lines g
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Simulated data and resultsSimulated data and results
3D raster terrain (displayed in f )
Altitude=210m Overlap=4%Velocity=60m/s
surfer) Sensor-ALS-50Firing frequency=20KHz Scan frequency=48HzScan angle=40°Flight area=430m 430m
Bharat Lohani, IIT Kanpur India
LiDAR data plot in plan
A-A
B-B
Bharat Lohani, IIT Kanpur India
Profile A-A without error
Profile A-A with error
Bharat Lohani, IIT Kanpur India
Profile B-B w. r. t. flight lines g
Bharat Lohani, IIT Kanpur India
LiDAR data without error
Bharat Lohani, IIT Kanpur India
LiDAR data with error
Bharat Lohani, IIT Kanpur India
Data with no attitude variation
Bharat Lohani, IIT Kanpur India
Data with attitude variation
Bharat Lohani, IIT Kanpur India
Fractal data displayed in surfer
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Terrain with objects
Bharat Lohani, IIT Kanpur India
LiDAR data of terrain with objects Altitude=490m Overlap=2%Velocity=60m/sSensor-ALS-50Firing frequency=20KHz Scan frequency=48HzScan angle=50°Flight area=640m×460m
Bharat Lohani, IIT Kanpur India
Altitude=400m Overlap=2%Velocity=60m/sSensor-ALS-50
Effect of data density Firing frequency=20KHz Scan frequency=48HzScan angle=50°
Bharat Lohani, IIT Kanpur India
Altitude=400m Overlap=2%Velocity=60m/sSensor-ALS-50Firing frequency=5 KHz Scan frequency=48HzScan angle=50°
Bharat Lohani, IIT Kanpur India
1
2Profile view of buildings
2
Profile view-1
Profile view-2
Bharat Lohani, IIT Kanpur India
Effect of different altitude Altitude=200m Overlap=2%Velocity=60m/sSensor-ALS-50Sensor-ALS-50Firing frequency=5 KHz Scan frequency=48HzScan angle=50°
Bharat Lohani, IIT Kanpur India
Altitude=100m Overlap=2%Velocity=60m/sSensor-ALS-50Sensor-ALS-50Firing frequency=5 KHz Scan frequency=48HzScan angle=50°
Bharat Lohani, IIT Kanpur India
Effect of different scan angle Altitude=200m Overlap=2%Velocity=60m/sSensor-ALS-50Sensor-ALS-50Firing frequency=5 KHz Scan frequency=48HzScan angle=50°
Bharat Lohani, IIT Kanpur India
Altitude=200m Altitude=200m Overlap=2%Velocity=60m/sSensor-ALS-50Firing frequency=5
Scan angle=32
Firing frequency 5 KHz Scan frequency=48HzScan angle=32°
Bharat Lohani, IIT Kanpur India
Effect of different flight direction
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Applications of simulatorApplications of simulator
Education
To understand:
Process of data generationEffect of change in various parameters on dataEffect of errors on data
Bharat Lohani, IIT Kanpur India
Laboratory exercises
Data with varied specificationsData with varied specifications
Full and accurate ground truth knownFull and accurate ground truth known
Bharat Lohani, IIT Kanpur India
Research projects
Evaluation of Information extraction algorithmsg
Assessing effect of error on performance of galgorithms
Finding optimal data specifications for an application
Bharat Lohani, IIT Kanpur India
Application in building identification research
Bharat Lohani, IIT Kanpur India
Variation of accuracy indices
Bharat Lohani, IIT Kanpur India
Variation of accuracy indices
Bharat Lohani, IIT Kanpur India
Final touches…in next three months
Multiple return implementationMultiple return implementation
Error introduction in individual parametersError introduction in individual parameters
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
Bharat Lohani, IIT Kanpur India
)())(2(cos))(2(sin ttk
K
ktj
J
jiX idmidDCidBAa +⎟
⎞⎜⎛ ∏
+⎟⎞
⎜⎛ ∏
= ∑∑
j=2, k=2
)())((cos))((sin11
ttkk
ktjj
jX idmidT
DCidT
BAa +⎟⎠
⎜⎝
+⎟⎠
⎜⎝
∑∑==
0.38B1
1.65A2
3.54A1
0.51C1
2.45B2
1
0.88D2
2.77D1
3.77C2
1000T
0.001dt
0.0m
1000T
Bharat Lohani, IIT Kanpur India
)())(2(cos))(2(sin ttk
K
ktj
J
jiX idmidDCidBAa +⎟
⎞⎜⎛ ∏
+⎟⎞
⎜⎛ ∏
= ∑∑
j=2, k=2
)())((cos))((sin11
ttkk
ktjj
jX idmidT
DCidT
BAa +⎟⎠
⎜⎝
+⎟⎠
⎜⎝
∑∑==
3.38B1
3.25A2
0.68A1
0.89C1
4.45B2
1
1.34D2
5.23D1
2.54C2
1000T
0.001dt
0.0m
1000T
Bharat Lohani, IIT Kanpur India
)())(2(cos))(2(sin ttk
K
ktj
J
jiX idmidDCidBAa +⎟
⎞⎜⎛ ∏
+⎟⎞
⎜⎛ ∏
= ∑∑
j=2, k=2
)())((cos))((sin11
ttkk
ktjj
jX idmidT
DCidT
BAa +⎟⎠
⎜⎝
+⎟⎠
⎜⎝
∑∑==
4.38B1
0.25A2
1.78A1
1.78C1
2.45B2
1
2.34D2
3.23D1
1.24C2
1000T
0.001dt
0.003m
1000T
Bharat Lohani, IIT Kanpur India
)())(2(cos))(2(sin ttk
K
ktj
J
jiX idmidDCidBAa +⎟
⎞⎜⎛ ∏
+⎟⎞
⎜⎛ ∏
= ∑∑
j=2, k=2
)())((cos))((sin11
ttkk
ktjj
jX idmidT
DCidT
BAa +⎟⎠
⎜⎝
+⎟⎠
⎜⎝
∑∑==
4.38B1
0.25A2
1.0A1
0.78C1
5.45B2
1
4.34D2
7.23D1
1.24C2
1000T
0.001dt
0.004m
1000T
Bharat Lohani, IIT Kanpur India
j=3, k=3)())(2(cos))(2(sin ttk
K
ktj
J
jiX idmidDCidBAa +⎟
⎞⎜⎛ ∏
+⎟⎞
⎜⎛ ∏
= ∑∑
0.65A2
2.75A1
)())((cos))((sin11
ttkk
ktjj
jX idmidT
DCidT
BAa +⎟⎠
⎜⎝
+⎟⎠
⎜⎝
∑∑==
B2 4.45
B1 2.38
A3 0.6
C 1 77
C1
B3
1.51
80
D1
C3
2.77
C2 1.77
0.35
m
D3
D2
0.0
3.38
100
T
dt
1000
0.006
Bharat Lohani, IIT Kanpur India
Comparison of data sets without RPY and with Roll only.Equation of the surface: Z=sin(X/10)-sin(XY/90)-60.Flight velocity: 60 m/s.Flight height: 60 m.Distance: 30 m.Firing frequency: 5000 Hz.Scan frequency: 48 Hz
a d t o o y
Scan frequency: 48 Hz.Scan angle: 50 deg.
Bharat Lohani, IIT Kanpur India
Comparison of data sets without RPY and with Pitch only
Equation of the surface: Z=sin(X/10)-sin(XY/90)-60.Flight velocity: 60 m/s.Flight height: 60 m.Distance: 30 m.Firing frequency: 5000 Hz.Scan frequency: 48 Hz.Scan angle: 50 deg.
Bharat Lohani, IIT Kanpur India
Comparison of data sets without RPY and with Yaw onlyEquation of the surface: Z=sin(X/10)-sin(XY/90)-60.Flight velocity: 60 m/s.Flight height: 60 m.Distance: 30 m.Firing frequency: 5000 Hz.Scan frequency: 48 Hz.Scan angle: 50 deg.
Bharat Lohani, IIT Kanpur India
Comparison of data sets with and without RPYEquation of the surface: Z=sin(X/10)-sin(XY/90)-60.Flight velocity: 60 m/s.Flight height: 60 m.Distance: 30 m.Firing frequency: 5000 Hz.Scan frequency: 48 Hz.Scan angle: 50 degScan angle: 50 deg.
Bharat Lohani, IIT Kanpur India
Comparison of data sets with lower and higher ax
Eq ation of the s rface AX+B +CZ+D 0Equation of the surface: AX+By+CZ+D=0.(A=0, B=0, C=0, D= -65)Flight velocity: 60 m/s.Flight height: 65 m.Distance: 60 m.Firing frequency: 5000 Hz.Scan frequency: 48 Hz.Scan frequency: 48 Hz.Scan angle: 50 deg.
Bharat Lohani, IIT Kanpur India
LiDAR data of fractal terrain Altitude=500m Overlap=2%Velocity=60m/sSensor-ALS-50Firing frequency=20KHz Scan frequency=48HzScan angle=50°Flight area=550×550km
A-A
Bharat Lohani, IIT Kanpur India
Profile A-A
P fil B BProfile B-B
Bharat Lohani, IIT Kanpur India