Self-service bike sharing systems experience often imbalance problems: in some stations there is a lack of bikes while in others there are too many bikes, leaving no empty rack for a user willing to park his bike. Imbalance problems can be partially settled by the actions of vehicles moving bikes during the night in order to prepare the forthcoming day. We speak then of static regulation, since bikes can be assumed to be not moving. These problems can also be partially countered by actions of the vehicles during the day. This is dynamic regulation and forms the topic of our present work. We build a precise theoretical framework for studying this dynamic problem and discuss the kind of impact vehicles moving bikes can have on the system. We prove moreover the inherent hardness of the dynamic regulation problem and present some ways to circumvent it by the use of heuristics. A pricing technique, not involving vehicles, is also proposed. An open-source and versatile simulator is also briefly described and used to compare the aforementioned methods.
