Overview of shared-bike repositioning optimization with artificial intelligence

Abstract

Abstract Rapid developments in artificial intelligence (AI) present unprecedented opportunities to enhance the operational and management performance of shared bikes. heuristic algorithms, supervised algorithms, unsupervised algorithms, and reinforcement learning (RL) in AI technologies enable the consideration of more possibilities in the bike repositioning problem (BRP), including addressing challenges such as large-scale bike sharing, real-time dynamic repositioning, and dynamic policy interaction with the environment. This paper provides an overview of research on bike-sharing repositioning utilizing AI techniques. The applications of Heuristic Search methods and Machine Learning, including RL for docked and dock-less shared bikes, are summarized based on dynamic and static environments, respectively. We provide a comprehensive analysis of the advanced development in AI-based BRP and review the application of AI technologies in obtaining scientifically repositioning strategies that effectively balance supply and demand conflicts. Moreover, this study delves into the constraints and potential advancements of AI methods for shared bike reallocation, offering valuable recommendations for future research.