Login

Project

#185 Incentivizing Participation in Peer-to-Peer Ride-Sharing Platform


Principal Investigator
Fei Fang
Status
Completed
Start Date
July 1, 2018
End Date
June 30, 2019
Research Type
Applied
Grant Type
Research
Grant Program
FAST Act - Mobility National (2016 - 2022)
Grant Cycle
2018 Mobility21 UTC
Visibility
Public

Abstract

Many residents in rural and suburban areas such as Lawrence County and several cities in Allegheny County rely extensively on private modes of transportation in daily commutes and other trips, and others without private vehicles lack access to their essential needs. Peer-to-peer ridesharing platforms that promote shared modes of transportation provides a promising opportunity to reduce the costs of travel and provide transportation alternatives. Such platforms connect commuters and enhance information sharing so that carpooling opportunities can be identified and leveraged. The key to the success of a peer-to-peer ridesharing platform is active participation of commuters. This research proposes to design incentive mechanisms to engage participants, to nurture and promote participation, and to improve the efficacy of the platform, with the ultimate objectives of creating a wide range of travel options that are consistent with participants’ preferences and system-wide objectives. This research will focus on providing a set of principled solutions to incentivize participation, including (1) providing service tools for participants, (2) tapping into the participants’ intrinsic motivation via non-monetary reward schemes, and (3) leveraging extrinsic monetary incentives to further motivate the participants. The solutions will be built upon machine learning, mathematical programming, and computational mechanism design. Through real-world data based simulations and the pilot deployment of the designed solutions on a peer-to-peer ridesharing platform, it will assess its impact on participation pattern, participants’ satisfaction and the overall efficacy of the platform in terms of various criteria, such as the number of trips completed, the total emission reduction, and the total value of the fulfilled shared rides. This research can contribute to the ongoing development of technological solutions to promote ride-sharing, reduce the costs of travel, mitigate transportation-induced emissions, and enhance access to urban areas.    
Description
Transportation provides access to essential needs—healthcare, education, jobs, and human services. However, rural and suburban areas such as Lawrence County and several cities in Allegheny County often have limited public transportation options. For example, current public transportation options in Lawrence County consist of daily bus trips to and from Pittsburgh and are highly utilized. But they are concentrated at morning and evening rush hours and lack flexibility. The development of ride-hailing platforms like Uber and Lyft, unfortunately, does not address the problem, due to the relatively small supply base and the very high resulting costs of transportation to common destinations such as downtown Pittsburgh. As a result, many residents rely extensively on private modes of transportation in daily commutes and other trips. This may lead to high costs of travel, negative environmental impact, added congestion costs, as well as limited mobility for households that do not own a vehicle or only one vehicle.

A promising opportunity to reduce the costs of travel and enhance mobility lies in peer-to-peer ridesharing platforms that better connect commuters and promote shared modes of transportation, for instance among employees of the same firm, like Liberty Mutual and UPMC, or residents in the same residential community, like Hulton Arbors in Penn Hills. In these platforms (often web-based and/or mobile-based platform), each user could specify travel needs, access the requests of fellow users, and propose ride-sharing options. Complementary to ride-hailing platforms, peer-to-peer ride-sharing platforms enhance information sharing and coordination among commuters so that carpooling opportunities among travelers going in the same direction and/or to the same destination can be identified and leveraged. 

The key to the success of a peer-to-peer ridesharing platform is active participation of commuters.This research proposes to design incentive mechanisms to engage participants, to nurture and reward participation, and to improve the efficacy of the platform. It will first propose a set of novel incentivizing models to enhance participation in peer-to-peer ridesharing platforms, including providing service tools for participants, tapping into the participants’ intrinsic motivation via non-monetary reward schemes, and leveraging extrinsic monetary incentives to motivate the participants. It will then provide principled solutions to implement the proposed mechanisms in the optimal way using tools from machine learning, mathematical programming, and computational mechanism design, and assess the various mechanisms considered on participants and rural/suburban transportation systems through computational simulation and pilot deployment.

This project complements another project led by co-PI Alexandre Jacquillat and Prof. Vibhanshu Abhishek from CMU’s Heinz College, and supported by Mobility21, which focuses on the development of a peer-to-peer ride-sharing transportation platform for rural/suburban communities and the deployment of this platform in Lawrence County through field experimentation. In contrast, this project proposes to rigorously evaluate the design options underlying the design of such a platform so that the ultimate transportation options and outcomes are most aligned with the participants’ preferences as well as with system-wide objectives, such as congestion mitigation and emissions reductions.

More specifically, the proposed research will comprise five steps:

1. Design service tools. Firstly, we will focus on designing service tools for participants. The research question we aim to answer is how to dynamically match the users so that their flexibility constraints can be satisfied, and the overall efficacy can be maximized. A user’s request for a ride will be satisfied only if it is matched to another user who can provide a ride. Most users have their own preferences and flexibility constraints. For example, a user may be willing to provide a ride from Lawrence County to downtown Pittsburgh, if he can depart between 7:55am-8:05am, and the delayed time for himself is less than 10 minutes. Given such complex constraints, searching for matches and tracking the dynamics of the posts on the platform can lead to a huge cognitive burden for the users. Further, a user’s unguided selection among several options that are indifferent to him can sometimes significantly reduce the overall efficacy of the platform, since there could be other users who are competing for the same options but have more strict constraints. We will model the problem as a dynamic matching problem and design algorithms to compute the optimal matching over time. This research will lead to a tool that automatically recommends compatible carpooling options or ride requests to the users to reduce their cognitive burden while enhancing the overall efficacy. We will also test and compare different platform policies, such as (i) a recommendation-based system, where the platform recommends matching options but the users make the final decisions, (ii) a top-down system, where the platform imposes matching decisions to the users, (iii) a dynamic top-down system, where the platform imposes matching decisions to the users but may change them over time, etc.

2. Design non-monetary reward scheme. Secondly, we will tap in the participants’ intrinsic motivation via non-monetary reward schemes. The research question is how to design the reward scheme to encourage more participation and more shared rides, assuming the participants’ elasticity - the extent to which they are willing to deviate from his normal path to participate -  is a function of rewards. Peer-to-peer ridesharing can take place even in the absence of monetary incentives. The participants may join the program due to various reasons. By offering free rides, a participant may hope for getting a free ride when he needs one in the future. For participants who are potentially connected in other social aspects, e.g., employees of the same firm or residents in the same residential community, providing and requesting rides lead to closer connections between members in the community. Moreover, participants may be intrinsically motivated by the social benefits of carpooling (e.g., emissions reduction, reduced congestion). The best way to motivate them, in this case, is to nurture their intrinsic motivation, by providing virtual reward (points), creating ranked lists and milestones, which will engage them in a healthy competitive environment. Periodically, a participant can further be provided the chance to win a trophy based on the points he has earned. We study the problem of optimal incentive design under a game theoretic setting, taking into account the fact that the participants have their own utility functions. We will formulate the problem as a Stackelberg game, or leader-follower game, where the leader (i.e., the platform manager) decides the reward scheme, and the followers (i.e., the drivers and riders) are the participants who choose to adjust their flexibility based on the reward scheme. Intuitively, offering a ride with a flexible time window that can be matched to many ride requests is more valuable and should be awarded more points. However, finding a concrete and effective reward scheme remains a computational challenge. We will design algorithms to find the best strategy for the leader in the game and compute the optimal reward scheme through solving a bi-level optimization problem. The scheme can then be used in peer-to-peer ridesharing platforms.

3. Design monetary incentives. Thirdly, we will leverage extrinsic monetary incentives to motivate the participants. The research question is how to determine suggested price of ridesharing opportunities to the users who are willing to offer or accept monetary incentives to the other user matched to them, so as to encourage participation on the platform. This problem arises when most users need monetary incentives. However, unlike ride-hailing platforms such as Uber and Lyft, the primary goal of peer-to-peer ridesharing platforms is to help identify and leverage carpooling opportunities instead of making money through commissions. The platform is more in the role of providing a suggested price than enforcing a payment. Therefore, the design of pricing mechanism is very different from commercial ride-sharing platforms. Further, the explicit flexibility constraints provided by the users lead to a challenge that does not exist in ride-hailing platforms. We will design algorithms to compute suggested prices, with desired properties such as envyfreeness and maximal efficiency, which are common criteria in mechanism design.

4. Data collection and analysis. In all of the three aforementioned aspects, an estimation of future ridesharing opportunities and requests based on available survey data and real data collected during the deployment of the platform can further enhance the solutions. Our collaborators from Lawrence County and Allegheny County will collect data about transportation needs and ridesharing opportunities through surveys and the deployment of the baseline version of the peer-to-peer ridesharing platform that will be developed in the ongoing project led by co-PI Jacquillat and Prof. Vibhanshu Abhishek. When the data is available, we will apply machine learning models to predict the ridesharing opportunities, which can be fed in to the proposed solution approaches, and to evaluate the solution approaches through real-world data based simulations.

5. Tool development and pilot deployment. We will convert the solution approaches into integrated tools to the peer-to-peer ridesharing platform that will be developed in the ongoing project. The tools include a service tool that automatically recommends compatible carpooling options or ride requests to the users, a tool that shows the virtual reward (points) for providing ridesharing opportunities or requesting rides, and a tool that shows suggested prices of the ridesharing opportunities to the users. We will seek feedback from our partners through the deployment of the baseline version of the platform in Fall 2018, and build an enhanced version with the new tools.

Overall, this project will augment the peer-to-peer ridesharing platforms, and provide an additional layer of technological solutions to enhance mobility in a rural community, and reduce the costs of travel associated with fuel expenditures, car ownership, traffic congestion, and emissions.
    
Timeline
This project will be organized in five phases:
1. Solution development I (July – October 2018): Research on designing service tools.
2.  Solution development II (September 2018 – February 2019): Research on non-monetary reward scheme.
3.  Solution development III (January 2019 – June 2019): Research on monetary incentives.
4. Data collection and analysis (March 2019 – June 2019): Analysis of collected survey data, and development of updated solutions. Evaluation of solution through real-world data based simulations.
5. Tool development and pilot deployment (March 2019 – June 2019): Development of tools integrated into the peer-to-peer ridesharing platform and pilot deployment of the enhanced version of the platform.
    
Deployment Plan
The data analytics, modeling, algorithmic design, and platform development will be led by Prof. Fei Fang from CMU’s School of Computer Science and Prof. Alexandre Jacquillat from CMU’s Heinz College. It will involve two Ph.D. students, Hoon Oh (from CMU’s Computer Science Department) and Weilong Wang (from CMU’s Department of Civil and Environmental Engineering), who will be responsible for the data analytics, modeling, and algorithmic development of the incentive design. This project will also leverage funding from another project for the app development and the implementation of our proposed mechanisms in the app (led by co-PI Jacquillat and Prof. Vibhanshu Abhishek). Here, in this project, we are focusing on the mechanism design part of the research. Computing expenses were budgeted to support and complement the platform development efforts.

The initial data collection, the platform deployment, and the pilot data collection will be managed by the Lawrence County and Allegheny County. On the initial data collection side, this involves engaging with employers and administering surveys to the population. On the pilot deployment side, this includes the marketing of the platform, the training of its users, and any ongoing management issue that needs to be addressed. The research team at CMU will provide technical support and strategic guidance regarding the deployment plan, and will be engaged in the data collection and analysis efforts for impact assessment.
    
Expected Accomplishments and Metrics
The overall objective of the proposed research is to augment the peer-to-peer ridesharing platform and provide better technological solutions to enhance transparency across travel options and impact travel behaviors in a way that enhances quality of life, reduce travel costs and externalities, and improves mobility within a population. The goal of this project is to provide a solid step towards the objective with principled research and demonstrate its potential impact with real-world data from the Lawrence County and Allegheny County. Specifically, this will be quantified through three major performance metrics:
1. Platform utilization: Adoption rate of the platform, and user engagement (e.g., number of users over time; average daily usage)
2. Travel impact: Effect of the platform on travel behaviors, e.g., ride-sharing frequency, number of cars on the road, estimates of travel cost reductions
3. User mobility satisfaction: Impact of the platform on population satisfaction with the surrounding mobility systems and accessibility options, measured through surveys and interviews.    

Individuals Involved

Email Name Affiliation Role Position
feifang@cmu.edu Fang, Fei SCS PI Faculty - Untenured, Tenure Track
ajacquil@andrew.cmu.edu Jacquillat, Alexandre Carnegie Mellon Heinz College Co-PI Faculty - Untenured, Tenure Track

Budget

Amount of UTC Funds Awarded
$90000.00
Total Project Budget (from all funding sources)
$90000.00

Documents

Type Name Uploaded
Data Management Plan Data_Management_Plan_v2.docx Jan. 14, 2018, 11:38 a.m.
Presentation Meeting0628.pptx Sept. 29, 2018, 10:16 p.m.
Progress Report 185_Progress_Report_2018-09-30 Sept. 29, 2018, 10:16 p.m.
Publication p2pWriteUp.pdf March 30, 2019, 3:38 p.m.
Presentation 20181130_SMCseminar.pdf March 30, 2019, 3:38 p.m.
Presentation 20190315_UTCFacultyMeeting.pptx March 30, 2019, 3:38 p.m.
Progress Report 185_Progress_Report_2019-03-30 March 30, 2019, 3:38 p.m.
Final Report FinalReport.pdf July 16, 2019, 4:53 a.m.
Publication Spatio-temporal pricing for ridesharing platforms March 21, 2021, 7:20 p.m.
Publication Dynamic trip-vehicle dispatch with scheduled and on-demand requests March 21, 2021, 7:21 p.m.
Publication Optimal trip-vehicle dispatch with multi-type requests March 21, 2021, 7:22 p.m.

Match Sources

No match sources!

Partners

Name Type
Lawrence County Deployment Partner Deployment Partner
Allegheny County Office of Children, Youth and Families Deployment Partner Deployment Partner
Hulton Arbors Deployment Partner Deployment Partner