Obtaining Parking Information Without Prior Knowledge In Underground Parking Lot

Knowing the specific floor and area where a vehicle is parked can greatly assist individuals in locating their vehicle in multi-level underground parking lots. While most existing methods use pre-collected maps to provide this location-related information, humans can discern these details even in unfamiliar environments. This paper introduces a human-like perception framework that enables vehicles to autonomously acquire such location-related information without any prior knowledge. The proposed framework improves vehicle dead reckoning (VDR) and scene text recognition (STR) algorithms specifically for parking lot environments. The data from VDR and STR are integrated into a space-related buffer. By applying a majority voting strategy in this buffer, the system can confidently determine the correct floor and area information. The resulting location data is made accessible on the user’s mobile phone, offering invaluable assistance in locating the vehicle.