Transcript Tracking Technologies - UCL Computer Science

Virtual Environments:
Tracking
© Houari Abdallahi, James Lawton, Deniz Ozsen, Christine Dubreu
Outline

Introduction
• History
• Framework for Suitability
• 6DOF, Inside-out, Outside-in

Different Technologies:
• Mechanical Systems
• Optical
• Magnetic
• Acoustic
• Inertial


Comparison of Technologies
Conclusions
• Effects of inaccurate Position Tracking
• Performance Requirements for VR Illusion
Position Tracking
The real time tracking of the position
and orientation of points in space.
Tracking the head to control view
generation and tracking a pointing
device for interaction with the virtual
environment.
History
The groma consists of stones hanging
from sticks set at right angles to one
another.
Distant objects could be marked out
against the position of the stones in a
horizontal plane.
Egyptian Groma
History
The first head-mounted display system.
The Sword of Damocles
Framework for Suitability
• Resolution and accuracy
• Responsiveness:
 Sample Rate
 Data Rate
 Update Rate
 Lag
• Robustness
• Registration
• Sociability
 Range of operation
 Fitness for Tracking Multiple Objects
Tracking Technologies
Surging (depth)
Swaying (width)
Heaving (length)
Rolling
Pitching
Yawing
Tracking Technologies
The main tracking technologies used are
mechanical, magnetic, acoustic, inertial
and optical.
Most can be classified as:
Outside-in
Inside-out
Mechanical Trackers
• First systems used
• Linkages attached to a fixed
point
• Relies on the known geometry
of solid linkages
• Typically measures joint angles
Mechanical Trackers
• Some example systems
Mechanical Trackers
Sword of Damocles
Evaluation
Advantages
Disadvantages
• Accurate
• Motion Restriction
• Reliable (low lag)
• Poor Sociability
• Force-feedback
• Mechanical Part Wear-out
• No Line of Sight or Magnetic
Interference Problems
Optical Trackers
Setup
• Detectors
• Light
• Inside/Out Outside/In
Fixed Transducer System
• Relies on Known Distance between Emitters and Sensors
• Most Common
• Outside-In & Inside-Out Technologies
Advances in Optical Tracking
HiBall-3100
• An inside-out optical system by 3rdTech
• Autocalibration
• 12’x12’ to 40’x40’ setups
• 2,000Hz tracker update
• 0.2mm, 0.01° resolution
• Redundancy avoids occlusions
ARTtrack2
• Made by A.R.T GmbH
• Several configurations
– Camera with built-in flash
– Separate flash (flash can work
through walls, if they transmit
IR)
– Stereo cameras
ARTtrack2
• Made by A.R.T GmbH
• Several configurations
– Camera with built-in flash
– Separate flash (flash can
work through walls, if they
transmit IR)
– Stereo cameras
ARTtrack2
• Made by A.R.T GmbH
• Several configurations
– Camera with built-in flash
– Separate flash (flash can work
through walls, if they transmit
IR)
– Stereo cameras
Pattern Recognition System
• Compares Known Patterns to
Sensed Patterns
• LED Array
• Symbols
• Real objects
Laser Ranging
• Light Transmitted onto Object
• Changes in Reflected Light
Sensed
• Example: Medicine
Optical Tracker Evaluation
Advantages
Disadvantages
• High Availability
• Complexity
• High Date Rates
• Expense
• Good Sociability
• Range Error
• No Magnetic Interference
Problems
• Weight
• High Accuracy
• Line of Sight
Mechanical Vs Optical Trackers
Mechanical
Optical
• No Line of Sight Problems
• Line of Sight Requirement
• Poor Sociability
• Potential High Sociability
• Cumbersome
• Not Necessarily Cumbersome
• No Environmental Error
• Possible Errors from Lighting
• Good Responsiveness
• Better Responsiveness
• Static Accuracy
• Variable Accuracy (Range)
Magnetic/Electromagnetic Trackers
•
•
Emitter: Device that generates a magnetic field
Sensor: Device that measures the surrounding magnetic field
•
Advantages
– Very low latency (~5ms)
– Unaffected by sensor occlusion by non-ferromagnetic objects
 Ability to track multiple users  promotes Sociability
•
Problems
– Sensitive to interference by magnetic fields from devices in the working area
– Distortion of magnetic field by metallic objects
– Can make up for this using digital filtering algorithms, however, this lowers the
data rate!
– Magnetic field diminishes with distance  limited range of operation
•
Two varieties
– AC Emitter (e.g. Polhemus)
– DC Emitter (e.g. Ascension)
 less distortion of magnetic field by metallic
objects  better!
Magnetic Trackers - Examples
• Ascension Flock of Birds
–
–
–
–
–
–
DC
Up to 4 sensors
Long range coverage easily added by using “Extended Range Transmitter”
Update Rate: up to 144Hz
Accuracy of position: 1.8mm RMS
Accuracy of orientation: 0.5° RMS
• Polhemus Fastrak
–
–
–
–
–
–
AC
Single transmitter, up to 4 sensors
Update rate: 120Hz (for single receiver)
Latency: 4ms
Accuracy of position: 1mm RMS
Accuracy of orientation: 0.15° RMS
Acoustic Trackers
•
Emitter:
Device that generates an acoustic wave of a
certain frequency (speaker)
– Usually ultrasonic (> 20kHz)
•
Sensor:
•
Advantages
•
Problems
Device that registers sound (microphone)
– Inexpensive
– Can get high data rates, and high accuracy
– Sensor occlusion
– Interference by acoustic noise, such as keys jingling, and echoes from walls
Two varieties
• Time-of-Flight (TOF) Trackers
•
– Low speed of sound  low data rate
– Speed of sound affected by environmental factors such as temperature, pressure
Phase Coherent (PC) Trackers
– Higher data rate
– Can get position wrong by multiples of wavelength when object moves too fast
Acoustic Trackers – Examples (1)
• Component of Intersense Trackers (e.g. IS 900)
– TOF tracking
– SoniDiscs, mounted on SoniStrips on ceiling, generate 40kHz
signals
– At the same time, “Tracked Station” starts counter
– When signal arrives at “Tracked Station”, counter stops
• Logitech Red Baron: ultrasonic headtracker
• Mattel Power Glove
Acoustic Trackers – Examples (2)
• In2Games Gametrak
– Accuracy: 1mm
– Working volume: 3m cube
Inertial Trackers
• Sensors mounted on object that measure relative changes
in position and orientation
• How it works
– Change in position: Accelerometer
– Change in orientation: Inclinometer
• Advantages
– Autonomous: tracker on object “knows” where it is
• Problems
– All measurements are relative  Cumulative ranging errors!
Inertial Trackers - Example
• Component of Intersense Trackers (IS 900)
– Combined with acoustic tracking
Comparison of Technologies (1)
Mechanical
• Good
Accuracy
and
Resolutio
n
• Good
• Decrease as
working volume
increases (multiple emitters
sensors)
• Good
Responsiveness
Optical
• Good
• Well suited to
real time
applications
Magnetic
• Good in small
working volume
Acoustic
• Good
• Accuracy affected by ferromagnetic objects
• Relatively low
data rates
• TOF : Good in
small ranges
• Filtering can
introduce lag
• PC : Responsiveness
unaffected by
range
Comparison of Technologies (2)
Mechanical
Robustness
Sociability
Optical
• Good
• Good
• Not sensitive
to environmentally induced
errors
• Some systems
affected by
ambient light
• Limited range
• Range-accuracy
(multiple E/S)
• Two systems
cannot occupate • Inside out systhe same
tems more fit for
volume
tracking multiple
objects
•Vulnerable to
occlusion
Magnetic
Acoustic
• Ferromagnetic
objects create
eddy currents that
cause ranging
errors
• TOF : Vulnerability to ranging
errors
• Small working
volume (eddy
currents increase
with field
strength)
• TOF : Small
effective volumes
• Multiple emitters
• Unaffected by
non ferromagnetic occlusions
• PC : Excellent.
High data rates
unaffected by
range
• PC : Large
working volume
• vulnerables to
occlusion
Comparison of Technologies (3)
• Mechanical
– Cumbersome, well suited to force feedback
– Successful applications in Telerobotics
• Optical
– Compromise between range and accuracy
– Successfully used in cockpits
• Magnetic
– Relatively inexpensive, most commonly used in current VR
research
– Successfully used in cockpits
• Acoustic
– Starting to appear in marketplace
Effects of inaccuracy
• The tracked object can appear to be somewhere it is not
– When the tracked object is a part of the body, the illusion of the
simulated space tends to break down
• If a position tracker reports inaccurate data, the user has to
construct a mental model of the surrounding space from
incompatible data.
– Conflict between perceived visual space and perceived
proprioceptive space
– As the visual information tends to dominate, users can
experiment motion sickness.
Visual-proprioceptive conflicts
• Contention between the observed position of a limb and its
felt position
– Mismatches between the computer generated image and the
vestibular system
– The user will adjust input values to match with the visual informations
• Lag in reported body movement
– The user will minimize rapid movements that accentuate the visualproprioceptive conflict
– This inhibits natural movements and can interfere with the
applications requiring naturalistic simulations
• Jitter or oscillation of the represented body part
– Strongly contribute to motion sickness
Performance requirements
• The performance should be studied in terms of perceptual
datas and motion dynamics
• A gap exists between perceptual understanding and
technical practice
• This gap is not a barrier to current developpement as the
user unconsciously ajusts the visual or proprioceptive
processes
• We must understand the performance requirements of the
human perceptive system
Summary
• We have discussed several position-tracking technologies
and established a framework for suitability. A VR application
should provide the following :
–
–
–
–
High data rates for accurate mapping without lag
High tolerance to environmentally induced errors
Consistent registration between physical and virtual environments
Good sociability so that multiple users can move freely
• All of the technologies display both strengths ans
weaknesses. The ultimate tracker will probably not be
developped from a single technology, but as a hybrid of
these technologies.