PRTs (Personal Rapid Transits) are compact, electrical, guided, and automated vehicles, as an emerging public transport. They are controlled in central fashion to transport dynamically-arriving passengers to the destinations. To operate PRTs effectively, they are not only to be dispatched to present passengers but also to be relocated in advance against expected future passengers, especially when demands are different across stations. In this study, we propose a novel online dispatching approach by taking into account the minimal decisions that are required indispensably to operate PRTs. In our approach, dispatching and relocation are accomplished as a result of successive minimal decisions, which makes PRT control simplified and robust, while dispatching and relocation are explicitly considered separately by the previous studies usually in complicated manner. We make use of scenarios for generating minimal-decision alternatives and evaluating the best one to cope with uncertainty in passenger arrivals. The proposed approach is validated against the previous studies by discrete-event simulation.