01630163

Component

In many cities, taxis continuously circulate in search of customers. Such dynamic search behavior consumes much road space, contributing to local traffic congestion and air pollution. To better understand movements of vacant taxis, several studies have examined taxi drivers’ movement patterns. However, topics such as dynamic passenger finding strategies, uncertainty and learning processes have still been scarcely addressed. This paper proposes a behavioral agent-based model to simulate taxi drivers’ dynamic passenger search behavior under uncertainty. The model emphasizes: (i) taxi drivers’ subjective utility of passenger finding strategies under uncertainty, (ii) information learning and updating processes. Numerical experiments are conducted to examine whether the formulated model exhibit the desired emergent properties. Results indicate that the formulated model, based on Bayesian learning under uncertainty indeed is capable of learning and dynamically improving taxi drivers’ search strategies.

This paper was sponsored by TRB committee AP060 Standing Committee on Paratransit.

01618707

17-03432

English

13p

2017

Transportation Research Board 96th Annual Meeting

Digital/other

Figures; References

Choice models; Decision making; Intelligent agents; Machine learning; Microsimulation; Passengers; Taxicab drivers; Uncertainty

Agent based models

Data and Information Technology; Planning and Forecasting; Public Transportation

Transportation Research Board Annual Meeting 2017 Paper #17-03432

TRIS, TRB, ATRI

Dec 8 2016 11:17AM