Full text
| |
| Author(s): |
Douglas Oliveira Santos
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2014-12-12 |
| Examining board members: |
Eduardo Candido Xavier;
Luidi Gelabert Simonett;
Cid Carvalho de Souza
|
| Advisor: | Eduardo Candido Xavier |
| Abstract | |
In this work, we present a framework for dynamic ride-sharing. The framework has a client-server architecture. The client is used by passengers to request rides and by drivers, including vehicle owners and taxi drivers, who are willing to share their vehicles in order to reduce costs or to earn money. The server needs to solve a dynamic optimization problem which is proved to be NP-Hard. The problem, called Ride-sharing Problem with Money Incentive (RSPMI), is modeled in the following manner: at each instant of time, there are a set of passengers needing to travel from a source to a destination point and a set of vehicles, each one having a source and a destination. Passengers have constraints that need to be considered, which are: an earliest departure time, a latest arrival time, the number of passengers that will travel together and the maximum value they are willing to pay for the ride. Vehicles can have an earliest departure time and a latest arrival time, as well. They also have a maximum capacity and a price per kilometer. The problem is to compute a route for each vehicle, with the goal of maximizing the number of attended requests and minimizing the total paid by passengers. RSPMI is a new problem in the literature, differing from others ride-sharing problems, because it is the only one that considers shared costs, having a constraint which allows people to set the maximum value for the ride. The main focus of the work is to develop methods that can solve the dynamic version of the RSPMI, in real time and large scale. The proposed method needs an heuristic to solve the static problem and an algorithm to solve the Many to Many Shortest Path Problem. We developed GRASP heuristics for the static problem and used Contraction Hierarchies to deal with the shortest path problem. Computational experiments were made to evaluate our method and heuristics. We used instances based on real data that simulates a day of taxis activity in the city of Sao Paulo. In our experiments, passengers paid, on average, almost 30% less than a private ride (AU) | |
| FAPESP's process: | 13/06746-4 - Dynamic Taxi Sharing and Ridesharing: A Framework and Algorithms for the Optimization Problem |
| Grantee: | Douglas Oliveira Santos |
| Support Opportunities: | Scholarships in Brazil - Master |