Parking Alert System
From CS219
Contents |
Project
Summary
Parking policies can be hazardous to your wallet. Even though you pay the parking meter for an alloted time, a second too late is enough for a parking officer to leave a citation on your vehicle. The Parking Alert System reduces the risk of getting your vehicle cited, or towed, by notifying the user when their vehicle is about to violate one of the parking policies that the car is subject to. The Parking Alert System does this by, first, classifying the user as driving the vehicle or parking. An adaptive learning algorithm takes in GPS readings, along with inertial measurements (accelerometer), from the phone, to determine whether the car is in motion or parked. In the event of parking the vehicle, the system will lookup the parking policies for the immediate location (GIS information) and, if the system confidence is not above a threshold, prompt the user to confirm the parking policies. A centralized server will maintain a safety margin for the vehicle, and, if the safety margin falls below a threshold, the sytem will SMS the user to check on the vehicle. [1]
Overview
Please follow the overview by following the flow chart
Parking Alert System has been developed for the Nokia N95. The Python phone application of the Parking Alert System includes a classifier that periodically samples GPS and accelerometer traces to classify if the user has parked the vehicle. If the phone application classifies the user as having parked their vehicle, the phone application makes a standard HTTP request to the server.
The request consist of the latest longitude, latitude, speed, and heading of the vehicle from the GPS module on the phone. The server accepts the given data and searches a PostGIS-ready PostgreSQL database for all the parking restrictions that apply. The server returns a text response that is then parsed on the phone.
The phone parses the given restrictions and notifies, or schedules for, the user when a parking restriction is violated, or is about to be violated.
