ITS World Congress 2014 A proactive route search method for an efficient city surveillance Osamu Masutani Denso IT Laboratory, Inc. 1 Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
Apr 22, 2015
1
ITS World Congress 2014
A proactive route search method for an efficient city surveillance
Osamu Masutani
Denso IT Laboratory, Inc.Copyright (C) 2014 DENSO IT
LABORATORY,INC.All Rights Reserved.
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
2
Summary
Background : City surveillance and FCD
Sensing coverage for city surveillance
Method 1 : Radar-assisted FCD
Method 2 : Proactive route guidance
Conclusion & future work
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
3
Background : ITS for Smart City
ITS has potential to be an important fabric for sustainable smart city
Power train with less environmental impact
Efficient traffic management
Alternative way to monitor urban environment
City surveillance by vehicle
Key feature for smart city management
Crowd sourcing is promising = FCD
Denso Technical Review https://www.denso.co.jp/ja/aboutdenso/technology/dtr/v16/files/14.pdf
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
4
Key performance indices for city surveillance
KPI for city surveillance by FCD :
Sensing quality (Accuracy)
Quality sensors
Sensing quantity (Coverage)
Number of sensors
Trajectory of sensors = routing
Number of satellites
Trajectory (orbit) of satellitesCoverage
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
5
Enhancement of coverage by environment sensor
Utilize environment sensor in vehicle
In some scenario, detection of surrounding vehicle virtually be able to increase data points of sensing
Camera, Radar, Lidar, etc..
Radar-assisted FCD
Denso Technical Review https://www.denso.co.jp/ja/aboutdenso/technology/dtr/v17/files/10.pdf
http://www.embedded.com/print/4011081
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
6
Enhancement of coverage by proactive routing
Route control can dispatch each floating car efficiently
By dispersing their routes in the city
Can floating car be controlled by center ? – Yes !
In-direct traffic control (Signs, navigation)
Fleet management
Automated car
Proactive route searchProbing Route control
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
7
Core simulation function is available as Metro traffic simulator
Evaluation environment
Traffic simulation with a simple setting
Simple micro simulation
Grid map, One way traffic
City surveillance target : Traffic (Volume, Speed)
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
8
Evaluation 1 : Rader Assisted FCD (R-FCD)
Adaptive cruise control system can measure surrounding vehicles
Traffic volume and relative speed can be estimated
Evaluation
How radar leverage data points compared to conventional FCD ?
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
Evaluation 1 : Result - observation coverage
R-FCD produce much higher coverage than conventional FCD
9
Link IDtime
Traffic Volume
Full observation
FCD (10%)
R-FCD (10%)
Example Link
timeVolu
me
Observed traffic volume ratio
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
Evaluation 1 : Result - Link travel time estimation
Higher link coverage yields higher accuracy of link travel time estimation
10
Link IDtime
Link Travel Time
Full observation
FCD (10%)
R-FCD (10%)
Example Link
timeTr
avel TIm
e
Estimation error
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
11
Evaluation 2 : Proactive route search
Each FC reserves route before it arrives
Find optimal route according to number of reservations
Reserve each link on the route
Implementation
Telematics-based navigation
Path prediction
Connected - automated driving
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
12
Evaluation 2 : Evaluation spec
Indices
Coverage
Travel time
Compared with conventional route search methods
Static (distance cost)
Reactive (current traffic)
Proactive (reservation)1000 2000 3000 4000 5000
0
5
10
15
20StaticReactiveProactive
Simulation Steps
# of
con
gest
ed li
nks
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
13
Evaluation 2 : Result - both indices are improved
Coverage is increased
Travel time is reduced Distance
Reactive
Proactive
Traffic Volume Link ID
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
Evaluation 2 : Result - Trade-off
Trade-off between coverage and travel time
In this case, traffic dispersion also carries reduction of traffic congestion
The effect of reduction of traffic congestion compensate extension of travel distance.
14
300 350 400 450 500 5500.4
0.41
0.42
0.43
0.44
0.45
0.46
0.47
0.48
0.49
Average Travel Time [sec]Co
vera
ge
Proac
tive
Distance
Reactive
Relation between coverage and average travel time
Better
Worse
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
15
Conclusion and Future work
Both of two active sensing methods can enhance FCD
Radar assisted FCD can dramatically increase data amount
Proactive route search can also increase sensing area coverage without extending travel time of each car
Future work
More realistic evaluation (real city, real traffic)
More dynamic situation (dynamic sensing target)
16
Appendix
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights
Reserved.
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
17
Appendix – Scalability : traffic demand
Proactive route search enhance more in congested time
Traffic demand [volume/hour] Traffic demand [volume/hour]
Covera
ge
Travel ti
me
[sec]
Copyright (C) 2014 DENSO IT LABORATORY,INC. All Rights Reserved.
18
Appendix – Scalability : trip length
Proactive route search can be applicable for short trip
3 4 5 6 7 8 9 10111213141516171819200
0.10.20.30.40.50.60.70.80.9
1
DistanceRealTimeTravelTimeRealTimeVolumePredicted
Map length [blocks]
Covera
ge