Separated or integrated? the optimal matching mode in ride-hailing platforms with autonomous vehicles

As Autonomous Vehicles (AVs) become increasingly integrated into ride-hailing services, mixed platforms offering both AVs and human-driven vehicles have emerged. However, not all passengers are willing to accept AV ride-hailing services (ARS), requiring platforms to balance service matching modes and pricing strategies in light of heterogeneous passenger acceptances. To address this, we develop a queueing-game-theoretic model that captures the interactions between passengers and drivers on a two-sided platform. We analyze two service matching modes and their associated pricing strategies: separated, where AVs and human-driven vehicles operate independently and passengers freely choose between them; and integrated, where both vehicle types are integrated and passengers are randomly matched. Our key findings are threefold. First, the platform may not reduce the AV proportion as mismatch cost rises under integrated mode; in some cases, it may exclusively admit AVs, especially when ARS penetration is low. Second, the integrated mode yields higher total user surplus (passengers and drivers combined) when ARS penetration is significant, while the separated mode may perform better under low ARS penetration. Third, although time-sensitive passengers tend to prefer integrated, platforms may still favor separated at low ARS penetration. Notably, the integrated mode can uniquely achieve a win–win–win outcome for passengers, drivers, and the platform when ARS penetration is high. Finally, we develop a series of model extensions, including a differentiated wage scheme in the integrated mode, a refined waiting-time formulation, different AV ownership structures, and a setting in which conservative passengers may reject AV assignments. These robustness checks confirm the validity of our main results and highlights the nuanced trade-offs in mode selection for mixed ride-hailing platforms.