BikeButler: A New App That Helps Seattle Cyclists Find Their Perfect Route
Researchers at the University of Washington (UW) have unveiled BikeButler, a demo web application designed to help cyclists in Seattle find personalized bike routes. The app was presented on April 17 at the Association for Computing Machinery Conference on Human Factors in Computing Systems in Barcelona.
How It Works
Users input an origin and destination, then adjust sliders for eight different attributes—such as speed limits and greenery—to generate route options.
Routes are displayed with segment-specific notes on how they align with user preferences.
Users can also view images of route segments to compare different options side-by-side, making it easier to choose the best path.
Data Sources & Technology
- Street layout and basic information are sourced from OpenStreetMap and government data sets.
- Subjective attributes, like greenery and pavement quality, are analyzed using a visual language model (VLM) applied to Google Street View images.
- VLM ratings agreed with human raters approximately 60% of the time.
Development & User Testing
The initial prototype was built after working closely with four participants. It was later tested with 16 participants, and the overall response was positive.
Users reported having contextual preferences, such as seeking safer routes for leisure rides and faster routes for commuting. This feedback highlights the app’s potential to adapt to different riding needs.
Limitations & Future Work
- Quality is constrained by the recency and accuracy of available data.
- The app is currently limited to Seattle, though it is designed to be expandable to other cities.
Future work may include features such as:
- Route dragging for custom adjustments
- An option for fewer turns
- Better quantification of intersection preferences
Authors & Funding
Lead author: Jared Hwang, UW doctoral student
Senior author: Jon Froehlich, UW professor
Additional authors: John S. O'Meara, Zeyu Wang, Jasmine Zhang
The project was supported by the National Science Foundation.