ParkSense: Automatic Parking Positioning by Leveraging In-Vehicle Magnetic Field Variation

Abstract

Drivers often park their vehicles without consciously (physically or mentally) noting down the parked location which makes it hard and inconvenient for the users to later locate their vehicles. Existing solutions either require users to explicitly note down the parked position on their mobile device by performing a certain action, such as pressing a button to turn on their global positioning system, or are erroneous and inaccurate. This paper attempts to build ParkSense, a system that allows a smart phone to accurately and automatically "sense," and later navigate to, the position at which the vehicle was parked. We propose to make use of the variations of magnetic fields and electromagnetic fields inside the vehicles to detect when a user stops and turns off his or her vehicle. Our evaluation with an actual implementation of the system on different mobile platforms and operating systems, tested on different mobile devices/phones and car models, shows the detection accuracy of more than 90%, confirming the feasibility of our approach.