Franois PEYRET, LCPC

1
Cooperative positioning in CVIS

• François PEYRET, LCPC
• francois.peyret_at_lcpc.fr
• Cooperative systems workshop and product launch
• 10 December 2008
• Berlin

2
Outline

• The importance of positioning in cooperative
  systems
• Position is needed everywhere
• Positioning is a very challenging task
• GNSS is absolutely central, but not enough
• Which positioning ?
• Absolute position is never used at the end of the
  chain, but map-matched position
• A position without integrity indicator is very
  risky and even dangerous
• Cooperative positioning for cooperative systems
• Cooperative use of the infrastructure in the
  positioning/map-matching process
• Infrastructure could be physical or virtual
  (digital)
• The choices of CVIS
• "EGNOS inside" the accuracy and integrity of
  EGNOS availability
• Hybrid positioning the best we can expect from
  data fusion for absolute positioning
• WLAN positioning a support to hybrid positioning
• Map-matching after hybrid positioning a reliable
  map-matched position with confidence
• Emap-matching an innovative step towards
  lane-level positioning
• Conclusion

3
The importance of positioning in cooperative
systems

• Position is needed everywhere
• Positioning is a very challenging task
• GNSS is absolutely central, but not enough

4
Position is needed everywhere

• All CVIS, SAFESPOT and COOPERS applications are
  based upon the knowledge of the vehicle(s)
  position, most of the time in real time
• The position is useless without the map
  information, the real interesting information is
  the estimation on the vehicle position on the map

5
Positioning is a very challenging task, GPS is
not enough

• The apparently easy positioning performed by the
  standard navigation systems is misleading
• In reality, the quality of service of standard
  GPS in constrained environments is generally
  quite poor, although the global performance is
  really impressive
• Quality of service
• Availability
• Accuracy
• Integrity

6
Which positioning ?

• Absolute position is never used at the end of the
  chain, but map-matched position
• A position without integrity indicator is very
  risky and even dangerous

7
Absolute position is never used at the end
You are here !...
Sensor data fusion
X, Y, Z, V
GNSS sensor
l, f, H
D
Proprioceptive sensor 1
da/dt
Proprioceptive sensor 2
8
Integrity mechanism
High accuracy but no integrity
Low accuracy but with integrity
20cm
10cm
Acccuracy (error RMS)
8 cm
18 cm
Integrity (probability to remain inside the
yellow circle)
100
99
9
Interest of integrity for ADAS
10
Cooperative positioning for cooperative systems

• Cooperative use of the infrastructure in the
  positioning/map-matching process
• Infrastructure could be physical or digital

11
Cooperative positioning for cooperative systems

• GNSS alone is not enough
• Absolute position is useless
• cooperation with the infrastructure is mandatory
• Cooperation can be
• Digital for the map-matching
• Physical and digital to help the basic
  positioning
• Use of geo-located landmarks in the environment
• Use of digital map information to constrain the
  data fusion (see Emap-matching further)

12
The choices of CVIS

• "EGNOS inside" the accuracy and integrity of
  EGNOS availability
• Hybrid positioning the best we can expect from
  data fusion for absolute positioning
• WLAN positioning a support to hybrid positioning
• Map-matching after hybrid positioning a reliable
  map-matched position with confidence indicator
• Emap-matching an innovative step towards
  lane-level positioning

13
General architecture of on-board POMA positioning
services
Optional local Reference Data
DLR
L1/L2 GPS Receiver
SAPOS Module
Internet access
Alcatel Office (Toulouse)
Alcatel
Time Stamp
Infrastructure based positioning modules (WLAN,
WSN)
EGNOS Receiver
Precise relative absolute Position, Error
Infrastructure based positioning
Geosat Data Flow
LCPC (SMI)
EGNOS Data Collection Translation
Hybrid PVT EMAP Computation
Hybrid PVT, Errors Pos.rel. to Map
Hybrid PVT, Integrity, Errors
EGNOS Messages
LCPC (Livic)
PVT, Integrity, Sat PV, UERE
UTC
Hybrid Position Computation
Map Matching
Position rel. to Map, Confidence
EGNOS/PVT Computation
Time stamped DR Sensor Data
Raw data
Router PC
Map Data
Sensor Module
Host PC
PPS Sync
L1 GPS-EGNOS Receiver
Time Stamp
DR Sensors
PPS sync. GPS Time
Sensors
Processed Data
Physical Interface
(PVT Position, Velocity, Time)
Sensor Data
Logical Interface
Computation Module within POMA
14
EGNOS inside
Optional local Reference Data
DLR
L1/L2 GPS Receiver
SAPOS Module
Internet access
Alcatel Office (Toulouse)
Alcatel
Time Stamp
Infrastructure based positioning modules (WLAN,
WSN)
EGNOS Receiver
Precise relative absolute Position, Error
Infrastructure based positioning
Geosat Data Flow
LCPC (SMI)
EGNOS Data Collection Translation
Hybrid PVT EMAP Computation
Hybrid PVT, Errors Pos.rel. to Map
Hybrid PVT, Integrity, Errors
EGNOS Messages
LCPC (Livic)
PVT, Integrity, Sat PV, UERE
UTC
Hybrid Position Computation
Map Matching
Position rel. to Map, Confidence
EGNOS/PVT Computation
Time stamped DR Sensor Data
Raw data
Router PC
Map Data
Sensor Module
Host PC
PPS Sync
L1 GPS Receiver
Time Stamp
DR Sensors
PPS sync. GPS Time
Sensors
Processed Data
Physical Interface
(PVT Position, Velocity, Time)
Sensor Data
Logical Interface
Computation Module within POMA
15
EGNOS inside architecture
Vehicle
Platform (Toulouse Site)
Broadcast of EGNOS message via mobile
telecommunication is needed for urban application
where the EGNOS satellite may be masked more than
60 of the time.
GPS Raw Data (Pseudo Range)
EGNOS Inside Receiver
Filtered Position Horizontal Protection Level
EGNOS Data (Corrections)
GPRS
Application
Egnos Inside Server
16
EGNOS inside availability and SIS integrity

• EGNOS brings
• Improved accuracy (differential, corrections)
• Integrity of the Signal In Space (SIS)
• EGNOS inside brings additionally
• The availability of the EGNOS signals, normally
  send by a geostationary satellite
• The POMA module provides
• The Kalman filtered GPSEGNOS solution of the
  U-Blox receiver
• A GPSEGNOS Least Mean Square solution, MOPS
  compatible, with an HPL (Horizontal Protection
  Limit) computation for integrity
• a GPS-only Least Mean Square solution

17
Hybrid positioning
Optional local Reference Data
DLR
L1/L2 GPS Receiver
SAPOS Module
Internet access
Alcatel Office (Toulouse)
Alcatel
Time Stamp
Infrastructure based positioning modules (WLAN,
WSN)
EGNOS Receiver
Precise relative absolute Position, Error
Infrastructure based positioning
Geosat Data Flow
LCPC (SMI)
EGNOS Data Collection Translation
Hybrid PVT EMAP Computation
Hybrid PVT, Errors Pos.rel. to Map
Hybrid PVT, Integrity, Errors
EGNOS Messages
LCPC (Livic)
PVT, Integrity, Sat PV, UERE
UTC
Hybrid Position Computation
Map Matching
Position rel. to Map, Confidence
EGNOS/PVT Computation
Time stamped DR Sensor Data
Raw data
Router PC
Map Data
Sensor Module
Host PC
PPS Sync
L1 GPS Receiver
Time Stamp
DR Sensors
PPS sync. GPS Time
Sensors
Processed Data
Physical Interface
(PVT Position, Velocity, Time)
Sensor Data
Logical Interface
Computation Module within POMA
18
Hybrid positioning Interactive Multiple Model
GPS Data
CT
Prediction for CA model
CV
CA
past ego position
Constant Acceleration Constant Turning
Constant Velocity
Simple linear models for specific dynamics
19
Hybrid positioning a continuous solution
Blue Hybrid solution
Green EGNOS only
20
WLAN positioning a support to hybrid positioning
Optional local Reference Data
DLR
L1/L2 GPS Receiver
SAPOS Module
Internet access
Alcatel Office (Toulouse)
Alcatel
EGNOS Receiver
WLAN-based position
Precise relative absolute Position, Error
Geosat Data Flow
LCPC (SMI)
EGNOS Data Collection Translation
Hybrid PVT EMAP Computation
WLAN On Board RSSI Measurements
L11
GPS Time Stamp
EGNOS Messages
LCPC (Livic)
PVT, Integrity, Sat PV, UERE
UTC
Hybrid Position Computation
Map Matching
RSSI Measurements at known locations
EGNOS/PVT Computation
Supply RSSI
(RSSI, Exact Loc.)
Time stamped DR Sensor Data
Raw data
Router PC
WLAN Infrastructure Position Algorithm
Calibration Module
Sensor Module
Host PC
PPS Sync
L1 GPS Receiver
Time Stamp
DR Sensors
Supply Position
External
Sensors
Processed Data
Physical Interface
(PVT Position, Velocity, Time)
Sensor Data
Logical Interface
Computation Module within POMA
21
WLAN positioning basic principles

• Two steps
• Off-line Calibration
• On-board Localization (fingerprinting)

WLAN localization System works as a backup system
to GPS in regions critical for GPS.
RSSI Measurer
WLAN Positioning
Vehicle Position
RSSI Received Signal Strength Intensity
22
WLAN-based preliminary tests results
251 points were computed, using 89 different
Access Points (APs)
Average number of APs / point 8 (after
filtering)
Performance without hybrid and map-matching
modules
ErrLoc_Mean 6 m ErrLoc_Std 7
m ErrLoc_RMS 9 m
23
Map-matching after hybrid positioning
Optional local Reference Data
DLR
L1/L2 GPS Receiver
SAPOS Module
Internet access
Alcatel Office (Toulouse)
Alcatel
Time Stamp
Infrastructure based positioning modules (WLAN,
WSN)
EGNOS Receiver
Precise relative absolute Position, Error
Infrastructure based positioning
Geosat Data Flow
LCPC (SMI)
EGNOS Data Collection Translation
Hybrid PVT EMAP Computation
Hybrid PVT, Errors Pos.rel. to Map
Hybrid PVT, Integrity, Errors
EGNOS Messages
LCPC (Livic)
PVT, Integrity, Sat PV, UERE
UTC
Hybrid Position Computation
Map Matching
Position rel. to Map, Confidence
EGNOS/PVT Computation
Time stamped DR Sensor Data

• map-matched candidates (up to 10)
• vehicle estimated pose (position heading)
• timestamp
• for every candidate
• - longitudinal accuracy
• - confidence (likelihood)

Raw data
Router PC
Map Data
Sensor Module
Host PC
PPS Sync
L1 GPS Receiver
Time Stamp
DR Sensors
PPS sync. GPS Time
Sensors
Processed Data
Physical Interface
(PVT Position, Velocity, Time)
Sensor Data
Logical Interface
Computation Module within POMA
24
MM Longitudinal accuracy and confidence

0.4
0.1
0.2
0.3
Confidence indicator attached to each segment
25
Emap-matching an innovative step towards
lane-level positioning
Optional local Reference Data
DLR
L1/L2 GPS Receiver
SAPOS Module
Internet access
Alcatel Office (Toulouse)
Alcatel
Time Stamp
Infrastructure based positioning modules (WLAN,
WSN)
EGNOS Receiver
Precise relative absolute Position, Error
Infrastructure based positioning
Geosat Data Flow
LCPC (SMI)
EGNOS Data Collection Translation
Hybrid PVT EMAP Computation
Hybrid PVT, Errors Pos.rel. to Map
Hybrid PVT, Integrity, Errors
EGNOS Messages
LCPC (Livic)
PVT, Integrity, Sat PV, UERE
UTC
Hybrid Position Computation
Map Matching
Position rel. to Map, Confidence
EGNOS/PVT Computation
Time stamped DR Sensor Data
Emap Data
Raw data
Router PC
Map Data
Sensor Module
Host PC
PPS Sync
L1 GPS Receiver
Time Stamp
DR Sensors
PPS sync. GPS Time
Sensors
Processed Data
Physical Interface
(PVT Position, Velocity, Time)
Sensor Data
Logical Interface
Computation Module within POMA
26
Emap-matching a unique process for positioning
and accurate map-matching
You are here !...
Sensor data fusion
X, Y, Z, V
GNSS sensor
l, f, H
D
Proprioceptive sensor 1
da/dt
Proprioceptive sensor 2
27
Emap-matching a unique process for positioning
and accurate map-matching
Sensor data fusion and map-matching
X, Y, Z, V
GNSS sensor
D
Proprioceptive sensor 1
da/dt
Enhanced digital road database Emap
Proprioceptive sensor 2
Position on the map
Relevant attributes of the road segment
You are here, on this segment, at this abscissa,
on this lane !...
28
Emap-matching particle filter using map
constraints
Particle generation
DR measurements
State propagation model
Correction 1 geometrical and topological
constraints
EGNOS measurement
Correction2 weights modification
Composite state vector
Corrected particle set
Corrected estimated state
29
Conclusion
30
Conclusion
Development status

• Next challenges
• lane-level positioning with Emap-matching and
  reliable EGNOS
• use of integrity indicators at the application
  level
• how to transform the present RD POMA module
  into a commercial product ???

31
Thanks for your attention

• www.cvisproject.org