This study explores how automated personal travel assistants can be used to understand what motivates current travel behaviors, as well as what efficiency gains are feasible through alternative mode choice. The Trip Itinerary Optimization (TrIO) platform was developed to optimize trip itineraries in accordance with realistic daily constraints and unique values of individual travelers. More than 270 participants from across the United States used this three-stage web tool. In the context of shared mobility services and other emerging technical innovations, personal mobility applications like the one proposed in this report will play increasingly important role for those living in urban environments.
The online mobility recommendation process consisted of three stages. Users first provide origin-destination pairs, calendar events and travel constraints (the need to transport passengers or children, baggage, etc.) through an online travel diary interface. Additionally, a preference elicitation survey is given to measure prioritization of optimal time, cost, calories, and carbon emission outcomes when making travel decisions. The travel itinerary informs a recommendation engine that schedules requests to location-based web services for information including traffic conditions, estimated drive times between origin-destination pairs, and the cost and availability of public transit options and shared mobility services. Later, after feasible travel itineraries have been compiled from these web-based information services, platform users are provided access to a personalized online dashboard displaying sets of daily trip itineraries, each optimized for either time, cost, calories, carbon emissions or user preference. A week later, users were invited to respond to a final survey about their actual travel behaviors. At all stages, users responded to survey questions designed to quantify travel attitudes and behaviors.
Overall, users agreed that the dashboard was enjoyable and easy to use in addition to being helpful in adopting new travel behaviors. Results indicated that users retained a strong preference for optimizing time and cost and were more likely to use these itineraries versus the caloric intensity and carbon emissions itineraries. Users commonly expressed interest in even more fine-grained traffic data and the ability to dynamically model the effect of changes in travel times.
Users were also generally concerned over the safety of walking and biking recommendations. Responses indicate that more information about the amenities available to cyclists and pedestrians (sidewalks, shade from trees, access to food) and routes that avoid areas of perceived elevated danger would reduce barriers to implementing such recommended modes.