1. Instant Social Ride-Sharing Győző Gidófalvi * Uppsala University, Dept. of Information Technology Gergely Herényi motoros.hu: Online Hungarian Forum for Mobility and Transport Torben Bach Pedersen * Aalborg University, Dept of Computer Science

2. November 19th, 20082008 ITS World Congress, NY, NY2 Outline  Ride-sharing background  Instant Social Ride-Sharing System / Service Process and system components Mobile demo application  Dual-objective grouping of trips into ride-shares Detour measure between offer and request Social connection between participants Calculating a ride-share assignment Modes of operation  Simulated data sets  Evaluation of transport effectiveness  Conclusions and future work

3. November 19th, 20082008 ITS World Congress, NY, NY3 Ride-Sharing Background  Congestion, parking, and pollution are increasing in most cities  A potential solution is ride-sharing (carpooling) Encouraged by measures:  exclusive lanes, parking places, and reduced road tolls for carpools But wide-spread adoption is hindered by barriers:  Lack of effective mechanisms for scheduling/coordinating ride-shares  Safety risks  Social discomfort in sharing private spaces  Imbalance of costs and benefits among parties  Current commercial products / system for ride-sharing: nuRide, Carpoolworld, liftShare, eRideShare, etc… Post and search trips Manually construct / negotiate regular ride-shares Trust managed through a self-regulatory user-rating  Existing systems do not effectively eliminate barriers!

4. November 19th, 20082008 ITS World Congress, NY, NY4 Instant Social Ride-Sharing System / Service  Idea: Share a ride with a friend or a friend of a friend! Exploit exponentially growing popularity of social networking and the voluntary sharing of personal information online to eliminate the social barrier to ride- sharing Utilize mobile technologies (communication, computing and positioning) to allow instant ride-sharing, automatic scheduling and to help coordinate ride-shares  Question: Are there enough ride-sharing possibilities between friends to allow for effective transport?

5. November 19th, 20082008 ITS World Congress, NY, NY5 SRSS Process and System Components

6. November 19th, 20082008 ITS World Congress, NY, NY6 SRSS Mobile Demo Application Web demo: http://www.motoros.hu/SRSS/SRSS.html Java mobile application: http://www.motoros.hu/SRSS/SRSS.jad

7. November 19th, 20082008 ITS World Congress, NY, NY7 Dual-Objective Grouping of Trips into Ride-Shares  Minimize the extent of the “detour” that offering parties must make in order to serve requests  Maximize the amount of social connections amongst participants of ride–shares  Expected to increase the social comfort level and trust among ride-share participants  Leads to increased user acceptance and adoption of ride–sharing

8. November 19th, 20082008 ITS World Congress, NY, NY8 Detour Measure Between Offer and Request  Break displacements between origin and destination of an offer and a request into perpendicular offset components  Relative detour : fraction of the weighted average of the offset components and the shared distance  Detour measure is calculated using an SQL function

9. November 19th, 20082008 ITS World Congress, NY, NY9 Social Connection Between Participants  Number of relatively short paths between two users indicates the strength of the social connection, ssc  ssc values between pairs of connected users are pre- calculated and stored in an RDBMS using a self-join

10. November 19th, 20082008 ITS World Congress, NY, NY10 Calculating a Ride-Share Assignment 1.Calculate candidate matches between active offers and valid requests such that: 1.The users of the offer and the request are socially connected 2.The offering user has to make a detour that is less than a maximum threshold: max_det 3.The seat supply matches seat demand 2.Derive a single match score that is linearly related to the required detour and inversely related to the ssc between the users 3.Iteratively, in a greedy fashion assign requests to offers such that: 1.As many as possible of requests are assigned to an offer 2.The total match score of the assignment is minimized  The Ride-Share Assignment algorithm is effectively implemented as a stored procedure in a RDBMS  Refer to the paper for implementation details

11. November 19th, 20082008 ITS World Congress, NY, NY11 Modes of Operation  Requests can be assigned to stationary or to mobile (already departed) offers  Offers can depart, in an eager fashion, immediately after the first assignment, or can wait, in a lazy fashion, for more assignments

12. November 19th, 20082008 ITS World Congress, NY, NY12 Simulated Data Sets  Transportation data set: ST-ACTS: Trip = locations of two consecutive activities 1.74 million, on average 2.9 km long trips of 550 thousand prospective users for the course of a day  Social network data set: People tend to make social connections with other people that they frequently interact with in physical space (friends at home, colleagues) A user is assigned12 random friends, and 12 home and 12 work friends that are live and work close to the user  social network with 11 million links  SRSS user data set: Users join SRSS by invitation in the order of the number of received invitations and the utility the user sees in joining

13. November 19th, 20082008 ITS World Congress, NY, NY13 Evaluation of Transport Effectiveness  Transport effectiveness for varying number of users and under various parameter settings (max_det and max_wt) are evaluated according to three measures: resource- effectiveness, time-effectiveness, and reliability.

14. November 19th, 20082008 ITS World Congress, NY, NY14 Evaluation of Transport Effectiveness (cont.)  For 60,000 users (10% of the population): AVO-level is raised from 1 to 1.61 Users wait 2.4 minutes for a ride–share Offers have to make a detour that is 8% of the shared distance 91% of the requests can be served  Corresponding savings during the course of a single workday in a city like Copenhagen: 32% of the normal transport cost, specifically:  176,000 vehicle–kilometers  14,000 liters of fuel  32.7 tonnes of CO2–emissions  The proposed SRSS is clearly effective!

15. November 19th, 20082008 ITS World Congress, NY, NY15 Conclusions and Future Work  Conclusions Developed an instant Social Ride-Sharing System/Service (SRSS) along with a mobile demo application Experiments on simulated data sets show that the proposed SRSS can provide highly effective transportation and is feasible  Future work Road network distance based detour measure Highly scalable implementation of SRSS using a data stream management system Alternative system architecture possibilities: distributed