Dynamic Travel Information – Personalized and Delivered to Your Cell Phone

Abstract

The policy of FDOT is to use "the Florida Advance Traveler Information System as the primary method to disseminate timely and important travel information to the public so that the public can make informed decisions regarding their travel plans."Travelers can subscribe to alerts for specific road segments, reasons for the alert, and days of week.However, well-defined "push" alerts can also subject the user to a high number of text messages or emails.Researchers found that a single subscription to 2 highways in Tampa Bay resulted in 6851 emails and many more text messages, sent to a single user, with an average of more than 16 emails per Friday, over 1.4 years.These messages also may be irrelevant to the user based on his real-time location (e.g., not outside the coverage area) and next intended destination.In addition, alerts received while the traveler is driving increase the risks of distracted driving.Overwhelming the travelers with irrelevant information will likely decrease its value and, therefore, its ability to influence travel behavior.If such systems automatically "pull" the information and deliver it intelligently at pertinent time and place, the usefulness of the information would have an impact on the user's travel behavior; and that is the premise of this research.With this information, people can make more informed decisions about their travel options such as seeking an alternate route, changing a departure time or changing mode.This project explored the development of technology that delivers dynamic, personalized traffic alerts only when the alert is relevant to the user's real-time location or predicted next destination and departure time.This system also sought to incorporate real-time transit information based on nearest transit stop.The approach was to use TRAC-IT, a software architecture supporting simultaneous travel behavior data collection and real-time location-based services for Global Positioning System (GPS)-enabled mobile phones, as the basis for developing several software applications to deliver predictive messaging.A Fast GPS Clustering algorithm was developed to identify Points-of-Interest (POIs) that users frequently visit.The Trip Segmentation algorithm developed by the research team separates raw GPS data into trips from POI to another.Destination and Departure Time Predictions module uses history of POIs and trip information to predict the user's next destination and time of departure to deliver timely pre-trip information.Path Prediction estimates the actual path a user will take to their next destination and identifies any active incidents they may encounter.Finally, real-time transit information from Hillsborough Area Regional Transit and real-time traffic incidents from the FL511 Application Programming Interface (API) were integrated with TRAC-IT to demonstrate the feasibility of personalized services using data from real-world traveler information systems.A prototype mobile application for the Android (Google, Inc.) platform was also implemented to demonstrate a method of delivering traffic alerts to a cell phone only when the user is traveling below a threshold speed.The message can also be delivered in audio format using the Android Text-to-Speech API so the traveler is not required to read while driving.Using predictive technology and based on real-time and historical travel patterns, this prototype demonstrated that providing personalized traveler information that has the potential to affect trip-making decisions is indeed feasible.Additional research needs were identified before full-scale deployment.Given the current capabilities of cell phones and expectation of more advancement in the industry, this promising software can potentially offer the traveling public the means of adapting to current traffic or weather situations and foster more use of public transportation.17.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 3 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.