UMD Poster - 2017 AVS
incenTrip: A Real-Time Data-Driven Decision-Support Toolkit for the Incentivization and Guidance of Shared, Electrified, and Automated Vehicles (SEAVs)
Technology at a Glance
Potential Collaborations - SEAVs
►Adoption
of automated vehicles, Electrified vehicles
►Shared-vehicle/ride Long-Term Incentives Destination: Work Preferred Arrival: 8am
program participation ►Travel
and driving behavior shifts
►Eco-routing Pre-Trip Incentives
re-route
Real-Time Big Data Hub
Incentive Optimizer
►Regional
Integrated Transportation Information System (RITIS) at University of Maryland (UMD)
guidance positive social influence.
Accurate behavior intent prediction
Multimodal 4%
►Stated-Preference Drive
Rideshare 3%
Seed saver
►Over
3,000 samples ► Behavioral response under incentives ►Attitudes towards rideshare and SEAVs ►Work with DFHV and LM on SEAV scenarios
73%
LEAF SAVER
TREE SAVER
FOREST SAVER
PLANET SAVER
AWS Cloud Deployment
Alternative 1 Alternative 2
Alternative 1 Alternative 2
Feasible Alternatives Dedicated Data API
►Gamification,
►Passive
SM-CA Data API
Predictive Dataset
SM-CA Cloud Database
Alternative 1 Alternative 2
Control Architecture (CA)
CATT RITIS Database
Key Features of Our Approach •
Real-time API of traffic incidents, events, volumes, speeds, and climate condition
AgBM Behavior Model
• • • • • •
A large-scale simulation model based on Navteq network for D.C. and Baltimore
High-fidelity traffic prediction Data Fusion Real-time events/incidents Real-time traffic monitoring Weather conditions Real-time traffic simulation
DTALite iOS SDK
MOVESLite
Error: ± 3 min.
140 120 80 60
Estimated Travel Time Experienced Travel Time
40 20 0
•
Actual User Data App Deployment
160
100
Android SDK
•
The Geo-Location Heat-Map for one Anonymous App User
tracking ►Rich behavior evolution from actual app users ►App pop-up questions for decision verification
System Model (SM)
A Data-Driven Model
Time (min.)
►“Congestion is a 10% phenomenon”, a small % change can lead to significant drop in energy consumption and congestion ►“Carrots, not sticks”, personalized/optimized incentivization, features game-type activities and membership levels for loyalty, balances monetary and non-monetary incentives, utilizes social recognition and influence in a user friendly interface. ►Advanced methodology for behavior research supported by surveys, focus groups, laboratory experiments, naturalistic driving platforms and real world test beds ►Real-time prediction of traffic and user intent, powered by AI and data-driven approach. Responsively re-adjust information and incentives to achieve maximum efficiency
Transit 2%
►Prosocial Gamification
https://ritis.org
guidance
Decision-Science Support
Walk/Bike 18%
Connections: To connect over 40,000 D.C. commuters ►RideAmigos: Collaborate on a Ridesharing platform at UMD campus ►Incentive Optimizer, ►DC Dept. of For Hire Vehicles: Incentivize the usage of Electrified Taxi Efficient Budget Allocation ►Local Motors: effectively incentivize the adoption/usage of its AV: Olli
for SEAVs ►Dynamic
Real-Time Incentives
►Commuter
►Alpha Testing, with 355 driver samples. 3% incentivized to Rideshare; 24% to more energyefficient means of transportation
Chenfeng Xiong1*, Lei Zhang1, Mehrdad Shahabi2, Yafeng Yin2 1. National Transportation Center, University of Maryland, College Park 2. Dept. of Civil and Environmental Engineering, University of Michigan * [email protected]; 301-405-9430
Predicted Travel Times matches Actual Experiences
incenTrip Deployment ►24/7
Monitoring and comprehensive situation awareness, e.g. incidents, Etc. ►Real-time guidance of routing, departure times, and ride-matching for SEAVs ►Quick and accurate analysis using actual user data as feedback
Technology at a Glance
Potential Collaborations - SEAVs
►Adoption
of automated vehicles, Electrified vehicles
►Shared-vehicle/ride Long-Term Incentives Destination: Work Preferred Arrival: 8am
program participation ►Travel
and driving behavior shifts
►Eco-routing Pre-Trip Incentives
re-route
Real-Time Big Data Hub
Incentive Optimizer
►Regional
Integrated Transportation Information System (RITIS) at University of Maryland (UMD)
guidance positive social influence.
Accurate behavior intent prediction
Multimodal 4%
►Stated-Preference Drive
Rideshare 3%
Seed saver
►Over
3,000 samples ► Behavioral response under incentives ►Attitudes towards rideshare and SEAVs ►Work with DFHV and LM on SEAV scenarios
73%
LEAF SAVER
TREE SAVER
FOREST SAVER
PLANET SAVER
AWS Cloud Deployment
Alternative 1 Alternative 2
Alternative 1 Alternative 2
Feasible Alternatives Dedicated Data API
►Gamification,
►Passive
SM-CA Data API
Predictive Dataset
SM-CA Cloud Database
Alternative 1 Alternative 2
Control Architecture (CA)
CATT RITIS Database
Key Features of Our Approach •
Real-time API of traffic incidents, events, volumes, speeds, and climate condition
AgBM Behavior Model
• • • • • •
A large-scale simulation model based on Navteq network for D.C. and Baltimore
High-fidelity traffic prediction Data Fusion Real-time events/incidents Real-time traffic monitoring Weather conditions Real-time traffic simulation
DTALite iOS SDK
MOVESLite
Error: ± 3 min.
140 120 80 60
Estimated Travel Time Experienced Travel Time
40 20 0
•
Actual User Data App Deployment
160
100
Android SDK
•
The Geo-Location Heat-Map for one Anonymous App User
tracking ►Rich behavior evolution from actual app users ►App pop-up questions for decision verification
System Model (SM)
A Data-Driven Model
Time (min.)
►“Congestion is a 10% phenomenon”, a small % change can lead to significant drop in energy consumption and congestion ►“Carrots, not sticks”, personalized/optimized incentivization, features game-type activities and membership levels for loyalty, balances monetary and non-monetary incentives, utilizes social recognition and influence in a user friendly interface. ►Advanced methodology for behavior research supported by surveys, focus groups, laboratory experiments, naturalistic driving platforms and real world test beds ►Real-time prediction of traffic and user intent, powered by AI and data-driven approach. Responsively re-adjust information and incentives to achieve maximum efficiency
Transit 2%
►Prosocial Gamification
https://ritis.org
guidance
Decision-Science Support
Walk/Bike 18%
Connections: To connect over 40,000 D.C. commuters ►RideAmigos: Collaborate on a Ridesharing platform at UMD campus ►Incentive Optimizer, ►DC Dept. of For Hire Vehicles: Incentivize the usage of Electrified Taxi Efficient Budget Allocation ►Local Motors: effectively incentivize the adoption/usage of its AV: Olli
for SEAVs ►Dynamic
Real-Time Incentives
►Commuter
►Alpha Testing, with 355 driver samples. 3% incentivized to Rideshare; 24% to more energyefficient means of transportation
Chenfeng Xiong1*, Lei Zhang1, Mehrdad Shahabi2, Yafeng Yin2 1. National Transportation Center, University of Maryland, College Park 2. Dept. of Civil and Environmental Engineering, University of Michigan * [email protected]; 301-405-9430
Predicted Travel Times matches Actual Experiences
incenTrip Deployment ►24/7
Monitoring and comprehensive situation awareness, e.g. incidents, Etc. ►Real-time guidance of routing, departure times, and ride-matching for SEAVs ►Quick and accurate analysis using actual user data as feedback
