Treffer: An enhanced approximate dynamic programming approach to on-demand ride-Pooling.

Ride-pooling services have been growing in popularity, increasing the need for efficient and effective operations. The main goal of ride-pooling services is to maximise the number of passengers served while limiting wait and delay times. However, factors such as the timing and volume of passenger requests, pick-up and drop-off locations, available vehicle capacity, and the trajectory to fulfil multiple requests introduce high degrees of uncertainty, creating challenges for ride-pooling operators. This study aims to expand the current state-of-the-art Approximate Dynamic Programming (ADP) approach for ride-pooling services, introduce key extensions, and perform a comparative analysis with the Neural Approximate Dynamic Programming (NeurADP) approach to optimise the efficiency and effectiveness of these services. Specifically, we develop an ADP approach that incorporates three important problem specifications: (i) pick-up and drop-off deadlines, (ii) vehicle rebalancing, and (iii) allowing more than two passengers in a vehicle. We conduct a detailed numerical study with the New York City taxi-cab dataset and a dataset of taxi-cab requests collected in the city of Chicago. We also provide a sensitivity analysis on key model parameters such as wait and delay times, passenger group sizes, and vehicle capacity. Our comparative analysis highlights the strengths and limitations of both ADP and NeurADP methodologies. Network density and road directionality are found to significantly impact the performance. NeurADP is found to be more efficient in learning value function approximations for larger and more complex problem settings than the ADP approach. However, in less complex cases, ADP is shown to outperform NeurADP. [ABSTRACT FROM AUTHOR]

Copyright of Transportmetrica B: Transport Dynamics is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)