Enabling Sustainable Point-to-Point Transport Service via Behavioral Game Theory

Funding Source: SMU Urban Institute (MOE Academic Research Fund (AcRF) Tier 1)

Principal Investigator: Associate Professor Shih-Fen Cheng (SMU)

Synopsis: In dense urban environments, Point-to-Point (P2P) transport services (including both taxis and ride-hailing services) play a crucial role in reducing private car usage. However, the inefficiency in matching drivers and passengers often results in undesirable congestion and wasted capacity. In the literature, researchers have proposed policies and mechanisms such as surge pricing and guidance systems to balance drivers' locations and passenger demands in real-time dynamically. However, to construct these mechanisms or policies, idealistic assumptions on how drivers would behave often have to be assumed; the departure of these assumptions from realities often contributes to the performance drop when these models are eventually adopted in practice.

This proposal aims to come up with a more realistic driver behavioral model based on real-world driver trajectories to predict how a driver would make operational choices. The proposed model is based on the well-known Cognitive Hierarchy (CH) model, where drivers interact with each other in a game-theoretic framework, yet with a limited "level of reasoning". Compared to the classical CH model, this proposal aims to make the following contributions:

• Extending the CH model to incorporate sequential decision-making process with large state space and uncertainty.
• Handling large-scale and noisy behavioral data.
• Applying the refined model to large-scale real-world use cases.