m. troger, c. fösleitner, j. seybold teleconsult austria gmbh, austria
DESCRIPTION
GNSS/INS Integration using a Tightly Coupled Approach for Navigation of People with Visual Impairments. M. Troger, C. Fösleitner, J. Seybold TeleConsult Austria GmbH, Austria European Navigation Conference 2013 23 rd – 25 th April 2013 Vienna, Austria. Motivation (1/2). - PowerPoint PPT PresentationTRANSCRIPT
GNSS/INS Integration using a Tightly Coupled Approach for Navigation of People
with Visual Impairments
M. Troger, C. Fösleitner, J. Seybold
TeleConsult Austria GmbH, Austria
European Navigation Conference 2013
23rd – 25th April 2013
Vienna, Austria
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria2
Motivation (1/2)
Navigation of pedestrian through GNSS
• Position is determined by means of GNSS
• Route to the destination point (track point) is calculated
• Manoeuvring list is generated
• Visually guidance to the track point through Smartphone screen
How to proceed, if the pedestrian is visually impaired?
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria3
Motivation (2/2)
Navigation of visually impaired people
• Determination of position
• Binaural sounds generatedo Depending on user’s position
o Depending on user’s heading
o Depending on position of track point
• Binaural sounds provided to user through stereo-headphone
• Requirements:o High accuracy and availability of position
o High availability of heading
Fusion between GNSS and INS sensors
• Different properties of both systems qualify them for combined use
• Kalman Filter optimal tool for fusion
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria4
Contents
Loosely vs. tightly coupling
Tightly coupled approach
• Parameters of the state
• Measurement updates
• Dynamic model
Dynamic test and results
Conclusion and future work
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria5
Loosely vs. tightly coupling
Loosely coupling Tightly coupling
Advantage of tightly coupling: GNSS measurements can be used if less than four satellites are visible!
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria6
Tightly coupled approach (1/4)
Parameters of the state vector
• ECEF position
• ECEF velocity
• Distance error caused by receiver clock error
• Velocity error caused by receiver clock drift
• Quaternion vector
• Biases of the gyroscopes
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria7
Tightly coupled approach (2/4)
Measurement update: GNSS observations
• Code pseudorangeso Corrected due to ionosphere, troposphere, and satellite clock error
o Elevation based measurement noise
• Dopplero Measurement noise indirect proportional to SNR
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria8
Tightly coupled approach (3/4)
Measurement update: Velocity
• Derivation of step frequency using autocorrelation of acceleration measurements
• Computation of horizontal velocity with step length and step length variation parameter
Measurement update: Bias of angular rates
• During phases without rotation and without movements
• Identification of phases using inertial measurements
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria9
Tightly coupled approach (4/4)
Time update
• Model of uniform motion
• Angular rates and gyro biases are introduced into dynamic coefficient matrix
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria10
Dynamic test and results (1/3)
• Positioning unit was fixed tightly onto the belt at the back of a test person
• Low-cost GPS L1 receiver with raw data outputo Code pseudoranges, Doppler (Output rate: 1 Hz)
o Navigation Message
• IMU with low-cost characteristicso Angular rates, accelerations (Output rate: 20 Hz)
• Synchronisation and storage of measurements on micro-processor board sensor fusion in post-processing
• Reference trajectory based only on GPS observations processed by means of Kalman filtering in real-time (1 Hz)
• Initial alignment at the beginning of the dynamic testo Roll and pitch from accelerations during static phase
o Heading from GPS during dynamic phase
• Default step length of user
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria11
Dynamic test and results (2/3)
Hea
ding
with
out
GP
S
outa
geH
eadi
ng w
ith s
imul
ated
G
PS
out
age
(35
s)
Hor
izon
tal p
ositi
ons
with
out
GP
S o
utag
eH
oriz
onta
l pos
ition
s w
ith
sim
ula
ted
GP
S o
utag
e (3
5 s
)
Dynamic test and results (3/3)
Overview of trajectories overlaid with GoogleTM Earth
Horizontal positions overlaid with GoogleTM Earth
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria13
Conclusion and future work
Conclusion
• Kalman Filter approach shows potential in area of pedestrian navigation
• GPS outages can be bridged over short time interval
• Heading information is usable for binaural guidance
Future work
• Find the optimal weighting between measurement noise and system noise
• Software integration for real time operation
• Online step length derivation
• Refinement of bias update
ENC 2013, April 23rd - 25th 2013 - Vienna, Austria14
Acknowledgement
The research work described was conducted within the project ARGUS, which is co-funded by the EU Seventh Framework Programme under grant FP7-288841(ARGUS).
E-MAIL [email protected]
WEB www.tca.at
PHONE +43-316-890971-11
FAX +43-316-890971-55
Markus TrogerSenior Systems Engineer