This project aims to optimize riders’ trips by adding a new feature (in-trip support/live updates) to the existing OpenTripPlanner (OTP), an open-source multimodal trip discovery and planning system trusted by major transit agencies nationwide. It will also make use of other existing open source technologies, including the General Transit Feed Specification (GTFS) and GTFS-real-time open data standards for transit schedule information and real-time updates for arrival/departure predictions and vehicle locations. The added new feature is expected to provide a complete, user-centric passenger experience by integrating trip planning, real-time updates, nearby stops and amenities, payment options, and other services. The OTP takes up open data, including static and real-time data for transit agencies, ride- hailing services, micromobility providers, bikeshare, and other first-/last-mile options. An OTP web app for desktops and mobile phones will be developed along with a kiosk UI optimized for touch screens and native OTP iOS and Android apps. The research team’s recently developed module, OTP-personas, will provide account management by which customers will be able to save favorite trips and stops, set frequented destinations, and receive service alerts through SMS, email, or push notifications.
The proposed enhancement to OTP will leverage datasets of current and archived conditions to simulate users’ planned trips against actual trip conditions. A machine learning algorithm (Kalman Filter or similar) will be applied throughout the iterative comparison of planned and actual trips to determine how much an actual trip’s conditions are deviating from the expected, scheduled trip conditions. This approach would give research team a confidence value for whether to give a user alternate directions/trip options. An appropriate confidence value would need to be determined to override a user’s planned trip with an alternate, rerouted itinerary.
With the addition of in-trip support feature and dynamic rerouting OTP enhancement, transit riders will be given accurate information about vehicle locations and how that may affect their planned trip. Once a traveler plans a trip, the location of the transit vehicle will be constantly updated and refreshed using high-quality GTFS-real-time data. Based on current conditions, historic information, and machine learning simulations, trips are reoptimized if transit vehicles are running behind schedule or if riders may miss planned transfers. The reoptimized trip can then be communicated to riders in real time through push notifications, SMS, or email.
Because OTP is an open-source project, the new feature will be made available for agencies using OTP to implement at no additional cost. Since in-trip support and dynamic re-routing not only benefit transit users but also the transit agencies, ttis enhancement has the potential to provide affordable and valuable insight to transit agencies to inform planning and operations. By analyzing common routes that are behind schedule or prevalent transfers missed by users, transit agencies have insight into operational deficiencies. Agencies can leverage data to modify schedules, synchronize transfers, plan for network redesign, and prioritize capital investments that improve transit reliability in problem areas. In the future, transit agencies could use in-trip support data to make real-time operation decisions.
The current version of General Bikeshare Feed Specification (GBFS) – v2.3 allows shared mobility providers to provide information about how long a vehicle can be reserved. This information could allow future extensions of this project to allow for in- trip support and dynamic rerouting that includes shared mobility vehicles, allowing OTP to recognize when a rider could reserve a shared vehicle to ensure its availability in the near future.
