Vehicle Tracking System in Underground Parking Lots Using Smartphone

Abstract

We propose surface correlation (SC), a technology for estimating the location of a vehicle in an underground parking lot in real time. The proposed technology is based on fingerprinting. The difference between the existing fingerprinting and SC is that the location is estimated using the received signal strength (RSS) vector sequence stored while the vehicle is moving. The distance moved by the vehicle is estimated using long short-term memory, and the direction is calculated using the gyroscope in a smartphone. Using the moving trajectory and accumulated RSS of the vehicle, a user RSS surface (URS) is generated, and the URS is compared with the radio map to find the location with the highest correlation. Additionally, the proposed technology provides a SC coefficient value (SCC), which is a correlation result between URS and radio maps. Through the SCC, the user can know whether the estimated location is reliable. To verify the performance of the proposed system, an experiment is conducted using 11 Bluetooth low-energy beacons in an actual underground parking lot, and an average accuracy of 3.2 m was obtained. It has been shown to outperform other algorithms, such as k-nearest neighbor and particle filters. This technology can provide accurate and robust localization performance using a smartphone in an underground parking lot, and it is expected that seamless navigation would be possible in various indoor spaces. Thus, the proposed technology will be used in the future to develop continuous navigation between vehicles and pedestrians.