Using Intelligent Collaborative Agents for Automating Distributed Taxi Dispatch

This paper presents the study of a novel approach towards an automated taxi dispatch system that handles current bookings in a distributed fashion. Existing systems in use by taxi operators in Singapore attempt to increase customer satisfaction locally, by sequentially dispatching nearby taxis to service customers. The proposed dispatch system attempts to increase customer satisfaction more globally, by concurrently dispatching multiple taxis to the same number of customers in the same geographical region, and vis-à-vis human driver and dispatcher satisfaction. To realize the system, we propose a multiagent architecture, populated with collaborative taxi agents that can actively negotiate on behalf of taxi drivers in groups of size N for available customer bookings. The dispatch and operational efficiency of the existing and proposed dispatch systems were evaluated through computer simulations on MITSIMLab, an existing simulation-based laboratory originally developed for evaluating traffic management system designs at the operational level. The empirical results, obtained for a 1000-strong taxi fleet over a discrete range of N , show that the proposed system can dispatch taxis up to over 50% faster, with up to 41.8% and 41.2% reduction in customer waiting time and empty taxi cruising time, respectively. A more efficient dispatch system can help maintain a higher standard of customer service vis-à-vis human driver and dispatcher satisfaction, especially when the demand for taxi service is manageable for the fleet size. Note to Practitioners— With the liberalization of the taxi industry in Singapore, keener competition among taxi operators has emerged. We believe the taxi operator that leads the competition will be the one with the best automated taxi dispatch system, offering the highest cost productivity and customer satisfaction. However, recent statistics on customer satisfaction, as reported in the news (e.g., [1]), question the efficiency of existing taxi dispatch systems. While many factors might have influenced and contributed to this outcome, our research is motivated by the need for better automated approaches to match customer service requests and taxis, whose arrival and availability, respectively, might be sporadic or not known a priori. In this paper, we propose a novel multiagent system, called NTuCab dispatch, to automate taxi dispatch in a distributed fashion. Our experiments for a 1000-strong taxi fleet show that NTuCab dispatch can outperform existing centralized dispatch, in terms of reduction in dispatch, customer waiting and empty taxi cruising times. Additionally, NTuCab dispatch can be implemented on an existing technological infrastructure, providing the opportunities to harness the existing power of multiple intelligent transportation technologies. In future research, prior to a largeK.T. Seow is with the Division of Computing Systems, School Computer Engineering, Nanyang Technological University, Republic of Singapore 639798. [email protected]. N.H. Dang is with the Division of Computing Systems, School Computer Engineering, Nanyang Technological University, Republic of Singapore 639798. [email protected]. D.H. Lee is with the Department of Civil Engineering, The National University of Singapore, Republic of Singapore 117576. [email protected]. scale implementation, we will need to add more features and investigate their effectiveness towards achieving overall efficiency, including techniques to influence and better match the physical distributivity between service demand and non-empty taxi supply in real-time.