VelObserver makes the quality of cycle routes visible from the public’s perspective. People rate individual route sections using photographs and their own experiences. Combining these assessments spatially helps reveal where improvements to the cycling network are particularly needed. Adrian F. Meyer works with POSMO Genossenschaft as Data Scientist and Lead AI Development, developing AI-supported analysis methods for cycling infrastructure. The project brings data analysis together with the experiences of people who use the cycling network in everyday life.
Context and significance
An infrastructure category does not fully describe whether cycling along a street feels safe or comfortable. Parked cars, mixed traffic and ambiguous road layouts can all affect the experience. VelObserver therefore connects formal infrastructure information with photographs and location-specific observations from people who cycle there.
The AI component is intended to structure visual evidence and make comparisons more systematic. Human assessments and machine-derived features nevertheless remain distinct: a model can identify potential infrastructure issues, but cannot on its own decide how a street feels to its users.
01
The research question
Where do the people who use the cycling network see the greatest need for improvement?
02
Finding an answer
- Bring together assessments of individual route sections based on photographs and everyday experiences.
- Connect spatially located user experiences with data analysis.
- Develop AI-supported methods to analyse cycling infrastructure.
03
Findings and applications
- Spatially aggregated assessments of cycle-route quality.
- A shared information base for developing the cycling network around users’ needs.
