Taxi behavior simulation and improvement with agent-based modeling

Abstract

Taxi services, despite deemed as a convenient form of commuting, are challenged by many issues. The issues can be categorized based on the perspectives of drivers and passengers. Regarding the issue from the driver’s perspective, taxi drivers are working longer hours. However, the revenue generated does not justify their increased working hours, (i.e. working longer hours with less revenue), which further implies that drivers are not getting enough passengers. By contrast, from the perspective of passengers, the prime issue with taxi services is that passengers are rejected or denied service. In this research, we aim to establish a taxi behavior simulation model for an existing conventional taxi operation and introduce optimization for this type of taxi operation. The evaluation between the two models shows, that by introducing optimization to the usual taxi behavior such as in providing greater flexibility in selecting a passenger, an improved service can be achieved for both taxi drivers and passengers.