Artificial Dataset Generation for Modeling and Simulation of Shared Electric Automated and Connected Mobility Systems with Autonomous Repositioning: A Survey

The introduction of Shared, Electric, Automated, and Connected Mobility (SEACM) systems will completely transform the transportation sector in the near future. However, these systems present additional challenges that require comprehensive performance analysis and effective deployment techniques. This paper will investigate the importance of creating artificial datasets for training Machine Learning models (ML), with a focus on generating datasets specific to SEACM systems. In contrast, to generate artificial datasets for the SEACM system, many characteristics should be considered such as user behavior, vehicle dynamics, charging infrastructure, and environmental conditions. As a result, using these synthetic datasets, decision-makers can correctly simulate several scenarios and evaluate the performance of these complex systems. This review paper also shows how artificial datasets may be used to train ML models to simulate SEACM with autonomous repositioning. This allows researchers to assess how well ML-based algorithms can be used to optimize vehicle routing, charging infrastructure usage, demand forecasting, and other important operational elements. In addition, we highlight the significant impact that the generation of artificial datasets has on performance evaluation and deployment approaches. Academics can develop efficient deployment strategies that maximize the effectiveness and sustainability of shared automated and connected electric mobility systems by training ML models on these datasets to better understand the strengths and limitations of various algorithms, identify potential areas for improvement, and more. Taken as a whole, this research highlights the importance of generating artificial datasets in the context of shared automated and connected electric mobility systems. It illustrates the methods used to generate these datasets, their importance for modeling and simulation, and their implications for performance evaluations and deployment strategies. To more rapidly build and optimize machine learning (ML) models for shared mobility systems, researchers and practitioners can exploit the potential of artificial datasets. This will ultimately lead to better transportation experiences and sustainable urban mobility.