One promising area that can be serviced by edge computing in real-time is autonomous
driving. Fully self-driving vehicles can operate on roads and in buildings, such as indoor parking lots, using
various sensors and communication modules. However, because the communication between indoor parking
lots and the outside world is limited, and autonomous vehicles currently lack the real-time performance
capabilities needed to process all information independently, it is necessary to develop a control scheme for
fully self-driving vehicles in indoor settings. In this study, we propose a smart parking lot for self-driving
vehicles based on edge cluster computing. A smart parking lot consists of fixed edges and mobile edge
vehicles and uses grid maps for parking lot management. To evaluate the performance of smart parking, we
compared parking time and moving distance in existing parking environments. Furthermore, the resource
cost and number of data transmissions were analyzed to confirm the number of edges for effective service
provision and maintenance.