E CO- FRIENDLY R EDUCTION OF T RAVEL T IMES IN E UROPEAN S MART C ITIES Daniel H. Stolfi [email protected]Enrique Alba [email protected]Departamento de Lenguajes y Ciencias de la Computación University of Malaga Genetic and Evolutionary Computation Conference July 2014 Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 1 / 20
21
Embed
Eco-friendly Reduction of Travel Times in European Smart Cities (GECCO'14)
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 2 / 20
IntroductionProposal
ExperimentsConclusions
INTRODUCTION
Nowadays there is a higher amount of vehicles in streets
The number of traffic jams is increasing
Tons of air pollutants are emitted to the atmosphere
The inhabitants’ health and quality of life is decreasing
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 3 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
RED SWARM
Our proposal, Red Swarm, consists of:A few spots distributed throughout the city
I Installed at traffic lightsI Linked to vehicles by using Wi-Fi
Our Evolutionary Algorithm
Our Rerouting AlgorithmSeveral User Terminal Units
I They visualize the alternatives routessuggested
I They could be smartphones, tablets, orOn Board Units
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 4 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
RED SWARM
Red Swarm offers:Personalized information for each vehicle (online, distributed)Prevention of traffic jamsReduction of greenhouse gas emissionsSensing of the city’s state
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 5 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
RED SWARM ARCHITECTURE
Configuration:Spot’s configuration is calculated by the Evolutionary Algorithm (offline)
Deployment and Use:Spots suggest new alternative routes to vehicles (online)
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 6 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
RED SWARM SPOT
Connects with vehicles and suggests alternative routes
Runs an instance of the Rerouting Algorithm
S1 and S2 are the Input Streets where vehicles arrive the junction
An output street is selected according to the probability value calculatedby our EA.
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 7 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
REROUTING EXAMPLE
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 8 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
SCENARIO BUILDING
We work with real maps imported from OpenStreetMapWe clean the irrelevant elements by using JOSM
We define the vehicle flows (experts’ solution) by using DUAROUTER
We import the city model into SUMO by using NETCONVERT
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 9 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
CASE STUDIES (I)
MalagaI 2.5 Km2
I 262 traffic lightsI 10 Red Swarm spotsI 1200 vehiclesI 169 routes
StockholmI 2.9 Km2
I 498 traffic lightsI 12 Red Swarm spotsI 1400 vehiclesI 131 routes
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 10 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
CASE STUDIES (II)
BerlinI 7 Km2
I 770 traffic lightsI 10 Red Swarm spotsI 1300 vehiclesI 122 routes
ParisI 5.6 Km2
I 575 traffic lightsI 10 Red Swarm spotsI 1200 vehiclesI 125 routes
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 11 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
SYSTEM CONFIGURATION
If a vehicle which is driving to Destination 2 enters by Street 1in the coverage area of a red swarm spot, a new route will besuggested by the Rerouting Algorithm according to theprobability values stored in the system configuration.
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 12 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
STATUS VECTOR
It represents the configuration of the N streets which are inputto a junction controlled by a red swarm spot. There are Mchunks of probabilities values in each street block in order tohold different configurations depending on the vehicles’ finaldestination.
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 13 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
EVOLUTIONARY ALGORITHM
The result of the algorithm is the configuration for allthe spots
The configuration is calculated in the offline stage.
(10+2)-EA
Evaluates individuals by using the SUMO trafficsimulator
The rerouting made by the Rerouting Algorithm isimplemented in SUMO by TraCI.
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 14 / 20
IntroductionProposal
ExperimentsConclusions
Red SwarmArchitectureCase StudiesEvolutionary Algorithm
FITNESS FUNCTION
F = α1(Θ − n) +
+ α21n
n∑i=1
COi + α31n
n∑i=1
CO2i + α41n
n∑i=1
HCi +
+ α51n
n∑i=1
PMi + α61n
n∑i=1
NOi + α71n
n∑i=1
Fueli (1)
Θ: Total amount of vehicles
n: Vehicles that end their itinerary during the period analyzed
α1 to α7: Normalize each variable
The lower, the better
Daniel H. Stolfi & Enrique Alba Eco-friendly Reduction of Travel Times. . . 15 / 20
IntroductionProposal
ExperimentsConclusions
Results50 ScenariosGraphs
AVERAGE AND BEST IMPROVEMENTS
We have reduced the CO, CO2, HC, PM, and NO emissions
We have also reduced travel times and fuel consumption
Case Study T .Time CO CO2 HC PM NO FuelMalaga 5.5% 4.1% -1.5% 3.0% 0.9% -1.8% -1.6%
Average Stockholm 14.2% 12.6% 3.2% 11.0% 8.5% 3.0% 3.0%50 Berlin 11.7% 10.6% 1.7% 8.7% 6.0% 1.5% 1.6%