Ride-sourcing services are now reshaping the way people travel by effectively connecting drivers and passengers through mobile internets. Online matching between idle drivers and waiting passengers is one of the most key components in a ride-sourcing system. The average pickup distance or time is an important measurement of system efficiency since it affects both passengers’ waiting time and drivers’ utilization rate. It is naturally expected that a more effective bipartite matching (with smaller average pickup time) can be implemented if the platform accumulates more idle drivers and waiting passengers in the matching pool. A specific passenger request can also benefit from a delayed matching since he/she may be matched with closer idle drivers after waiting for a few seconds. Motivated by the potential benefits of delayed matching, this paper establishes a two-stage framework which incorporates a combinatorial optimization and multi-agent deep reinforcement learning methods. The multi-agent reinforcement learning methods are used to dynamically determine the delayed time for each passenger request (or the time at which each request enters the matching pool), while the combinatorial optimization conducts an optimal bipartite matching between idle drivers and waiting passengers in the matching pool. Four tailored reinforcement learning methods, delayed multi-agent deep Q learning (Delayed-M-DQN), delayed multi-agent actor-critic (Delayed-M-A2C), delayed multi-agent Proximal Policy Optimization (Delayed-M-PPO), and delayed multi-agent actor-critic with experience replay (Delayed-M-ACER), are developed. Through extensive empirical experiments with a well-designed simulator, we show that the proposed framework is able to remarkably improve system performances, by well balancing the trade-off among pick-up time, matching time, successful matching rate.

Shared autonomous vehicles (SAVs) are rapidly emerging as a viable alternative form of public transportation with the potential to provide adequate and user-friendly, on-demand services without having vehicle ownership. It has been argued that SAVs could revolutionize transportation systems and our current way of life. Although SAVs are likely to be introduced in developed countries first, there is little doubt that they would also have a significant effect and enormous market in developing nations. This study aimed to investigate the factors that influence public acceptance of SAVs, as well as the current public attitude toward SAVs, in two developing countries, namely, Pakistan and China. A stated preference survey was conducted to understand respondents’ travel patterns, preferences, and sociodemographic data. A total of 910 valid responses were gathered: 551 from Lahore, Pakistan, and 359 from Dalian, China. A multinomial logit model and a mixed multinomial logit model with panel effect were used for data analysis. The results suggested that generic attributes, such as respondents’ waiting time, travel time, and travel cost were found to be significant in both cities. The results indicate that sociodemographic characteristics, such as education, income, travel frequency in a week, and people who had driver’s licenses, are significantly correlated with respondents’ interest in using SAV in Lahore. The results also showed that people who had a private car indicated a greater interest in SAVs in Dalian. The study provides a new perspective to understand the public preferences toward SAVs in developing countries with different economies and cultures, as well as a benchmark for policymakers to make effective policies for the future implementation of SAVs.

The introduction of vehicle automation, shared mobility, and vehicle electrification will bring about changes in urban transportation, land use, energy, and the environment. The accurate estimation of these effects is therefore essential for sustainable urban development. However, existing research on estimating the energy and environmental effects of shared autonomous electric vehicles generally ignores the interaction between land-use and transportation systems. This study, therefore, analyzes the long-term effects of shared autonomous vehicles (SAVs) from the perspective of land use and transportation integration. Different SAV pricing scenarios are also developed to explore the optimal pricing strategy for low carbon–oriented SAVs. Moreover, the study has further assessed the effect of vehicle electrification on vehicle emissions and energy consumption. The results have shown a nonlinear relationship between SAV fares and their transportation, land-use, energy, and environmental effects. Under an appropriate pricing strategy, SAV deployment could reduce PM_2.5 emission and energy consumption by 56–64% and 53–61%, respectively. With the further introduction of vehicle electrification, these can rise to 76% and 74%.

Solving the demand prediction problem is an important part of improving the efficiency and reliability of ride-hailing services. Spatial-temporal graph learning methods have shown potential in modeling the spatial-temporal dependencies of ride-hailing demand data, but most existing studies focus on region-level demand prediction with only a few researchers addressing the problem of origin-destination (OD) demand prediction. In addition, previous spatial-temporal graph learning methods employ pre-defined and rigid graph structures that do not reveal the instinct and dynamic dependencies of ride-hailing demand data. In this paper, we propose a joint learning framework called Dynamic Auto-structuring Graph Neural Network (DAGNN) to address the origin-destination demand prediction problem. We develop a Dynamic Graph Decomposition and Recombination layer (DGDR) to handle both the graph structure and the graph representation learning problems simultaneously, with graph representations learned from a group of trainable and time-aware edge-induced subgraphs. Experimental results show that our proposed model outperforms ten baseline models with two real-world ride-hailing demand datasets and is efficient in structural pattern discovery. Comparing with existing methods, the significant advantage of the proposed method is that it circumvents the difficulties in defining the underlying graph structure of the researched data.