Transcript Document

From Trajectories of Moving Objects to
Route-Based Traffic Prediction and
Management
Developing a Benchmark for Using Trajectories of
Moving Objects in Traffic Prediction and Management
by
Gyozo Gidofalvi
Ehsan Saqib
Presented by Bo Mao
2010-09-14
MPA'10 (GIScience 2010)
1
Traffic problems
Road network
Adoption of GPS based movement
Work
Pred. or act.
traffic event
Renewable
pseudo ID
MOD
Home
Location anonymization
Route-Based Traffic Prediction and
Management
Route-Based Traffic Prediction
and Management Server
Steam of
Evolving
Traj.
Relevant
Traffic
Info
Recent
Traj.
Traffic
Mngt.
Unit
Traj.
Mining
Unit
Traj.
Pred.
Unit
Frequent
Routes
Frequent
Route
Knowledge
Bank
k-anonymity
Frequent routes are explicit
inference units
Example traffic prediction and management tasks:




Estimate current/future traffic flow
Predict the near-future locations of vehicles
Which vehicles to inform in case of an event?
How and which vehicles to re-route in case of an event?
2010-09-14
MPA'10 (GIScience 2010)
2
Trajectory Data
 Number of objects: 1500 taxis and 400 trucks
 Measuring technology: GPS (+ accelerometer) based
measuring position (+ speed and heading)
 Location sampling: every 60 sec for taxies with passengers
(off-route less frequently) and every 30 sec for trucks
 Area/extent: Greater Stockholm area approximately 100km
by 100km
 Data rate/size: 170 million measurements per year / 1000
measurements per minute
 Availability: provided by Trafik Stockholm and is available at
the Transport and Logistic Division of the Department of
Urban Planning and Environment, Royal Institute of
Technology (KTH), Sweden
2010-09-14
MPA'10 (GIScience 2010)
3
Trajectory Data (2)
Measurements for 100 vehicles for a day
2010-09-14
Raw trajectories for 10 vehicles for a day
MPA'10 (GIScience 2010)
4
Traffic Management Benchmark
 Need to design a benchmark to evaluate the performance,
accuracy and scalability of a proposed traffic management
system.
 Design considerations:
 Trajectory sample bias: taxis are special
 Absence of individual mobility patterns: methods relying on such
patterns cannot be meaningfully evaluated
 Need for privacy: evaluation under different privacy requirements
 Realistic scalability tests: simple duplication of data does not
increase spatial-temporal density of it and is thus unrealistic
2010-09-14
MPA'10 (GIScience 2010)
5
Mobility Patterns
Frequent routes (speed + flow) for a day
2010-09-14
Speed deviations from the daily norm at 8am
MPA'10 (GIScience 2010)
6