Precise Tracking for Geospatial Augmented Reality

1
Precise Tracking forGeospatial Augmented Reality

• Gerhard Schall
• Institute for Computer Graphics and Vision
• Graz University of Technology

2
Agenda

• Introduction
• Modeling Pipeline
• Geospatial 3D Models
• Handheld AR-Setup
• Referece Systems
• Position Tracking
• Orientation Tracking
• Vidente Project

3
Geospatial Augmented Reality

• provide users with geospatial information, in a
  mobile
• context, adapted to their needs and their
  location
• Key issues
• Dynamic 3D modeling
• Location Based Services and real time geospatial
  data delivery
• Acquisition and spatial reference
• Multi-modal cognition and visualisation
• Visualisation of urban 3D models registered in 3D

4
Modeling Pipeline

• Exploiting data from GeoDBMS (e.g. cadastral
  survey,
• utility asset management) for real-world 3D
  content

5
Geospatial 3D Models

• Indoor Outdoor

6
Handheld AR-Setup

• Hardware
• USB Hub
• GPS receiver
• Interial Tracker
• Camera
• Weight minimizations
• Telescope antenna for GPS
• receiver
• Software
• GPS module configurable
• Gauss-Krüger projection
• UTM projection
• Settings for WAAS accuracy
• Playback feature

7
Reference Systems
8
Position Tracking

• Highly configurable Ublox Antaris 4 GPS receiver
• Current accuracy in our AR apps of 2 - 5m

9
APOS (Austrian POsitioning Service)

• APOS Real TimeDGPS
• submeter accuracy for outdoor AR positioning (30
  cm)

10
Orientation Tracking

• RMS Angular Resolution 1 in yaw, 0.25 in
  pitch and roll
• Update Rate 180 Hz
• Latency 2 ms

11
Tracking Results

• Collected tracking data of Inertia tracker and
  GPS
• Orientation
• FTW orientation sensor battery, BT, not shielded
• Inertial Tracker InertiaCube3 (IC3)
• IC3 has low drift
• IC3 less disturbed by magn. Fields
• Position
• Evaluation of 2 GPS receivers (Antaris Ublox,
  Holux GPSlim 236 BT)
• Antaris Ublox highly configurable
• Using SBAS correction information
• Visual Tracking
• Algorithms implemented in Hardware (FPGA) could
  greatly speed up tracking and leave GPU power for
  visualisation

12
Vidente Project

• stands for egocentric visualization of
• subsurface infrastructure data from Smallworld,
• providing a variety of dynamic visualization
  styles

http//studierstube.icg.tu-graz.ac.at/vidente/
13
Vidente

• user with the handheld device seeing
• the superimposed urban AR model

14
Conclusion and Outlook

• Handheld Geospatial Augmented Reality
• Geospatial urban 3D models from real-world data
• Precise Tracking -gt high quality registration -gt
  Better user experience
• Handhelds ongoing integration of sensors
• Camera
• GPS
• Orientation sensor
• etc.
• Ability to create great user experiences

15

• Thanks for listening!

16

• Questions?