Published on Sep 03, 2020
Monitoring road and traffic conditions in a city is a problem widely studied. Several methods have been proposed towards addressing this problem. Several proposed techniques require dedicated hardware such as GPS devices and accelerometrs
in vehicles or cameras on roadside and near traffic signals.
All such methods are expensive in terms of monetary cost and human effort required. We propose Wolverine1 - a non-intrusive method that uses sensors present on smartphones.
We extend a prior study to improve the algorithm based on using accelerometer, GPS and magnetometer sensor readings for traffic and road conditions detection. We are specifically interested in identifying braking events frequent braking indicates congested traffic conditions - and bumps on the roads to characterize the type of road. We evaluate the effectiveness of the proposed method based on experiments conducted on the roads in Mumbai, with promising results.
Several methods have been proposed that use sensors in smartphones for activity detection in various environments(Indoor localization, traffic detection and detecting activity of a person. The smartphone based traffic estimation methods obviate the need for specialized hardware installed in vehicles or on the road side.
These crowdsourced solutions(using distributed participatory data collection) have the advantage of high scalability as the number of smartphone users is growing at a rapid pace. The Nericell system uses accelerometer, microphone, GSM Radio and GPS sensors available in smartphones that users carries with them. In a smartphone based method, the orientation of the phone could be arbitrary with respect to the direction of motion, and could also change repeatedly.
Hence, it is required to virtually reorient the axes of the phone with respect to the vehicle. Nericell uses accelerometer and GPS readings alone for this. The direction of gravity is used to sense the vertical orientation, and the acceleration recorded during a braking event is used to compute the horizontal orientation.
Autowitness, a system to track stolen property also uses an idea similar to Nericell in order to reorient the axes. Further, Nericell detects road and traffic conditions based on threshold based heuristics.
Wolverine is a method which is similar to the Nericell system in that it too uses smartphone sensors for traffic state monitoring. However, for axes reorientation, we use the magnetometer to find the horizontal orientation of the phone instead
of waiting for a braking event.
This makes the system more reliable, and also reduces the energy intensive GPS usage. We give an energy consumption model for Wolverine, and compare it to Nericell, showing significant decrease in energy consumption, thus prologing the battery life.
Also, instead of threshold based heuristics for determining the traffic and road conditions, we use machine learning techniques (K-means clustering and Support Vector Machine (SVM)) which are more robust and versatile as compared to threshold based methods.
B.Tech Project by Ravi Bhoraskar