Vous êtes sur la page 1sur 76

HOSTED by

Aerial Mapping & Modeling


using UAV
Jeff Campbell & Mark Paulson
7 Mar 2015

www.verticalaspect.com

support@verticalaspect.com

Agenda

Our Background
Mapping Theory / Techniques / Products
UAV Image Capture
BREAK
Flight Demo
Post-Flight Processing
Wrap-up
2

Our Background
Jeff Campbell
Mark Paulson
Vertical Aspect

Jeff Campbell

Video production - KTVT-TV Dallas TX (1976-1978)


Naval Officer / Aviator (1978 2002)
(3000+ hrs. rotary and fixed wing Squadron CO)
H-60 Helicopter lead Systems Engineer Acquisition Professional
CIO of Navy Pentagon Staff

Lockheed Martin Corp

Program Manager for 4 IT programs (2002-2014)


Education

BA - Telecommunications Texas Tech (1975)

MS Univ of MD IT Resources Management (2000)


Dept. of Defense CIO Certification (2000)
R/C Aircraft

Hobbyist since 2011


Blade RC Helis, DJI F450 / F550 / S800 / P2V+ / Inspire 1
Formed Vertical Aspect with Mark Paulson (2014)
4

Mark Paulson.
Surveying / engineering / mapping (since 1968)
Registered Professional Land Surveyor (state of
Texas) since 1979
Involved in GIS (Geographic Information System)
since the mid-1980s
Involved in sales / training of SenseFly eBee fixed
wing UAV since Nov 2013
Formed Vertical Aspect 2014
5

Vertical Aspect
Services
Aerial mapping
Aerial Imagery

Pix4D Software
Premium Reseller
Training

UAV Training
6

Mapping Theory / Techniques /


Products
7

Graphical Map
BENEFITS:
Good for navigation
Inexpensive to produce
Easy to interpret
Non-technical
No underlying data
Not to scale / no accurate
measurements

Orthomosaic Map
BENEFITS:
Referenced to real world location
Enables accurate measurements
Rectified to 3D information
Graphical representation of
underlying data
Underlying data easily exported /
converted

Changing the mapping game


UAV photogrammetry
Efficient for Small areas
Affordable / Accurate maps

Survey efficiency with respect to Survey area

10

Mercator Projection

11

Lambert Conformal Projection

12

Map Projection Systems

13

Geolocation
Ground Control Points (GCP)
Well distributed in dataset
Visible on multiple images
5-8 sufficient for up to 1000 images
Quality report shows reprojection error
Obtaining GCPs
Traditional Survey Methods
RTK GPS device
Less accurate
Google Earth
Smartphone App
Handheld GPS device
14

Without GCP

15

With GCP

16

Mapping Output & Potential Market


Output
Orthomosaic
3D Point Cloud
Topographic
Digital Surface Model

Potential Customers
17

Output types

https://support.pix4d.com/hc/enus/articles/202561399-Example-Datasets-Availablefor-Download-Cadastre18

3D Models
Digital Surface Model
Everything

Digital Terrain Model


Bare Earth
Used by engineers

Oblique Models
19

Unclassified 3D Point Cloud

20

Classified Point Cloud-Terrain

21

Classified Point Cloud - Objects

22

Digital Surface Model (with scale)

23

Contour Lines

24

Contour Lines (terrain only)

25

Planar
(Photo Stitching)

Works only if terrain perfectly


flat
Only small datasets supported
(due to error accumulation)
No good georeference without
GCP
Distances (measurements) not
accurate
Requires only low number of
matches / keypoints (>100)

vs.

Orthomosaic
(Orthorectification)

Handles any terrain type


Handles large datasets
Supports Georeference
perfectly
Preserves distances &
becomes measurable
Requires high number of
matches/keypoints (>1000)
to generate the 3D model
26

Accurate Measurements

27

Stockpile Volume Calculations

28

Questions so far?

29

UAV Image Capture


Fixed Wing & Multirotor

30

Workflow
Plan & Load
the mission

Fly Capture
images

Process,
Provide Output

31

Fixed Wing vs. Multirotor


Fixed Wing
1. Larger area
2. Larger takeoff /
landing area
3. Higher Altitudes / less
detail
4. Longer Flight Time
5. More expensive

Multirotor
1. Smaller area
2. VTOL much smaller
takeoff / landing area
3. Lower altitudes /
greater detail
4. Shorter Flight Times
5. Lower cost of entry
32

Fixed Wing vs.

Multirotor

33

Methods of Capturing Images


Fixed Wing
SenseFly eBee
eMotion flight planning / capture

Multirotor
DJI Phantom Vision 2 +
Pix4D Capture App
34

Image Overlap
High accuracy results
Requires high overlap
between images
Consistent altitude
Recommended Overlap
Frontal - >75%

Frontal overlap > 75%

Side overlap > 60%

(with respect to flight


direction)

Side - >60%

(between flight tracks)

35

Camera Considerations
Megapixel size less important than image
overlap
Sensor size determines area of coverage at
given altitude
Automated Flight Path necessary for reliability
& accuracy
36

Safety Considerations

Autonomous Flight Required for proper data collection


DOs

Fly within Line of Sight


Remain below 400 AGL and outside of controlled airspace
Use spotter to handle people, have them keep their distance during takeoff and landing
Be prepared to take control of aircraft if autonomous flight fails
Give yourself a cushion on battery level

BE PREPARED
For the possibility of the aircraft losing power
Fixed wing more options & less weight than multirotor

DO NOT
Fly linearly over major roads

Cross them at right angles

Allow people to distract you during flight


Fly over crowds of people
37

Sample Fixed Wing Capture


(SenseFly eBee)

38

UAVDirect Mapping Flight

39

Todays Demo Flight

40

Questions so far?

41

Sample Multirotor Capture


(DJI Phantom Vision 2 +)

42

Requirements Phantom Mapping

Android or iOS Smartphone / tablet


DJI Phantom 2 Vision / Vision + System
Pix4D Capture App
PC with Pix4Dmapper Discovery (Free) or Pro

43

Pix4D Capture App


Download Pix4D Capture App (iOS or Android)
Choose mission type (Grid or Free flight)
Set up mission parameters

44

Pre-Takeoff

Start App
Connect to P2V+
Ensure Checklist complete
Takeoff - capture images autonomously

45

Takeoff / Flight

Flies pattern
Captures images autonomously
Returns to takeoff point
Downloads images to tablet

46

Todays Demo Flight

47

Questions so far?

48

HOSTED by

Aerial Mapping & Modeling

BREAK / DEMO FLIGHT


49

49

Post Flight Processing

50

Image Transfer
Transfer to tablet
Annotates images with more
accurate GPS geotags
Creates .p4d file

Upload to PC for processing


Pix4D cloud (initial quality
check) & later retrieval)
Dropbox
Manual transfer
51

Field Reporting (Optional)

52

Initial Processing

53

Quality Report

54

Initial Point Cloud

55

Ray Cloud

56

Densified Point Cloud

57

Mosaic Mesh Initial

58

Mosaic Mesh Reset

59

Mosaic - Blended

60

Contour and DSM

61

Google Earth & Pix4D

Google Earth

Pix4D Overlay

62

Capture App Settings

Higher altitude = Greater coverage / less detail


3 meters/sec
50 meters altitude

Camera Angle
90 (Vertical) Directly towards ground)

Camera Settings (use DJI Vision App)

Image size L
JPEG
ISO Auto or low value
White Balance Auto
EV 0

Reformat SD card prior to each flight


Better performance

63

Capture App Tips


App is not an autopilot, does not avoid obstacles
Be prepared to take control of P2V
(S1 switch in down position)

Refrain from manipulating app during flight


Both iOS and Android apps still in Beta
iOS more stable
Android currently has more features
Rotate pattern
Free Form Mission

Upcoming features
Selectable altitude (set at 50m)

Reformat SD card prior to each flight


64

iOS App
Slightly behind Android in development
Somewhat more stable (both Apps currently in Beta)

65

App Download

https://pix4d.com/download/

66

Other Uses

67

67

3D Modeling

44 Images
Canon EOS-7D
68

68

3D Modeling - GoPro
94 Images
GoPro Hero Black 3+

69

Pix4D & Accident Reconstruction

Typical accident reconstruction uses LIDAR or ground measurements


Aerial imaging offers significant improvements
Faster documentation & roadway re-opening

70

Highly accurate measurements

Truck Height

Truck Length

Truck Width

Tape Measure

1.70m

2.87m

2.43m

Laser Scanner

1.67m

2.87m

2.42m

UAV + Pix4D

1.70m

2.87m

2.43m

71

Inspire & Pix4D


Inspire Capture App under development
Awaiting DJI SDK

Vertical Aspect testing to date


Manually flying pattern
GPS tracker for more precise image geo-tagging

Inspire system shows great promise


Better lens / higher resolution
More robust platform
72

Recommended Hardware
Recommended:
Windows 7, 8 (64 bit).
CPU quad-core or hexa-core Intel i7/Xeon.
GPU (supports CUDA) NVIDIA GTX 750, GTX 970. 4GB RAM
Hard disk: SSD
No. of Images

RAM

Free HD Space

<100

8 GB

15 GB

100 to 500

16 GB

30 GB

500 to 2000

32 GB

60 GB
73

Pix4D Processing

Pix4Dmapper Discovery (free)

Pix4Dmapper Pro (free trial)

Orthomosaic, DSM
Elevation model
Point Cloud
Contour lines

https://pix4d.com/support/

Monthly rental, annual license, or one-time purchase

Quality Report
3D video animation (fly-through)

$350 / $3500 / $8700

Use our sample UAV images to process on your own


Pix4D Knowledge Base is free and accessible
Topic-specific tutorials

www.Pix4D.com
74

Pix4D Versions / Pricing

HANDS ON Pix4D TRAINING INCLUDED


With each One-Time Purchase,
Vertical Aspect will provide a full day of hands-on training*
75

Questions?
Jeff Campbell

Mark Paulson

www.verticalaspect.com
or e-mail us at

support@verticalaspect.com

76

Vous aimerez peut-être aussi