Tweeting Traffic Image Reports on the Road Daehan Kwak, Daeyoung Kim, Ruilin Liu, Liviu Iftode, Badri Nath Department of Computer Science, Rutgers University, Piscataway, NJ 08854-8091 {kwakno1, daeyoung.kim, rl475, iftode, badri}@ cs.rutgers.edu Abstract—As mobile phones have the ability to act as participatory sensors, we are beginning to witness the popularity of crowdsourcing and the sharing of traffic reports to improve the quality of the driving experience. This paper presents the architecture and implementation of a system called Social Vehicular Navigation (SVN), which allows users to generate and share geo-tagged image traffic reports called NaviTweets. Based on these traffic reports, Traffic Digests (concise snapshot summaries on the route of interest) are delivered to drivers to provide rich and reliable information supporting the route choice. These digests will complement factors such as the estimated travel time and assist the driver on their route choice decision making. The paper presents the initial design, along with a prototype implementation running on the Android platform, and details a user study conducted to evaluate the influence of providing traffic images on route-choice behavior. Keywords-social networks; vehicular networks; navigation systems/applications; route choice; traffic images I. INTRODUCTION With the ever-expanding affordability of cars throughout the world, traffic congestion is a severe problem that can have a negative impact on the economy, the environment, and human sentiment. There are a myriad ongoing attempts to alleviate traffic congestion, for example the infrastructure-based Intelligent Transportation Systems (ITS) such as on-ramp flow meters, traffic cameras, number plate recognition systems, and electronic informational displays (e.g. Variable Message Sign) along the roadways. Today, the most widely used solutions are infrastructure-less, named so because they use floating car data to determine traffic speed and to identify traffic congestion. Provided to the onboard navigation systems, this information can be used to calculate the fastest route. The infrastructure-less (or crowdsourcing) approach can be classified into two types: push and pull. The majority of today’s systems anonymously pulls GPS-based speed and location information from the client’s mobile or navigation system to provide real-time traffic information. The push type is based on user participation, where they push traffic reports in a richer context (e.g. the location of red light cameras and speed traps, the degree of traffic congestion, images, etc.) onto the server to share with other users. Waze [1] is an example of a navigation app that anonymously pulls speed and location information while providing an interface for drivers to push traffic reports. Traditionally, traffic reporting has been done primarily by the police, state departments of transportation, drivers reporting by phone, and also by traffic reporting companies. Such information is aggregated and then either resold or redistributed directly to the public, broadcast on-air by radio and TV stations, or used as traffic data for in-car navigation systems. Live traffic status reports are becoming more common and easily accessible, with traffic congestion maps available via online maps, mobile phones, and GPS devices. Nowadays, participation by the public in providing traffic reports is becoming popular, because it is easy for users to report and share traffic information with one another via smartphones. Traffic information can influence drivers’ route choice behavior, and consequently, guide them to less congested routes. When planning a route, current technologies collect and use real- time traffic information to calculate the route and then present a recommended list of alternative routes (normally two or three options) based on factors such as the Estimated Time of Arrival (ETA) or the shortest distance. Based on the list of optional routes, it is up to the driver to decide which route to take. Because ETA is the main factor that can be used in route decision, the design of vehicle navigation systems does not take into account other semantically richer information to support decision making and satisfaction of route selection. This paper focuses on how to provide a secondary level of detail using crowdsourced traffic images to support drivers in the selection of routes. To do so, we propose a system called Social Vehicular Navigation (SVN), which allows users to share image traffic reports, called NaviTweets. The users’ shared traffic reports are geo-tagged onto a map, called Social Traffic Map (a map representation). Based on many NaviTweets, a Traffic Digest is used to summarize the information of the route that is of interest to the user. Once the Traffic Digest is received, the information is displayed in a user-friendly way to the drivers to assist them with route selection. We explain the functions of the proposed SVN model in abstraction layers. Also, the system design for the SVN prototype implemented on the Android platform is presented along with results from a questionnaire survey to evaluate the usage of traffic images in route choice. This paper is organized as follows. The next section provides preliminaries by presenting related work. The proposed model for SVN is discussed in section III. The SVN system design and its prototype implementation is presented section IV and V, respectively. In section VI, the results of a user study based on a questionnaire survey is presented. Section VII presents a discussion and suggested future work and finally the conclusion of this paper in section VIII. This work was supported in part by the National Science Foundation (NSF) grant CNS-1111811. MobiCASE 2014, November 06-07, Austin, United States Copyright © 2014 ICST DOI 10.4108/icst.mobicase.2014.257815