Conclusion

Most people don’t feel safe walking at night in the dark — but current mapping and navigation software doesn’t take lighting into account when suggesting walking routes. Our model was created to help solve this problem. By optimizing for both travel distance and streetlight coverage, it can generate safer, more comfortable routes than traditional apps.

Our test results have been successful and match our expectations. As shown in the visualizations above, adjusting the balance between distance and lighting gives users flexible control over their route. We believe this model could be extended to any city or region with data available from OpenStreetMap and local lighting datasets.

That said, our project still has a few limitations:

• Currently, the model must start and end at intersections (nodes), not specific buildings or GPS coordinates. In the future, it would be useful to automatically choose the closest node based on a user’s actual location.

• The model only considers streetlamps, even though building lights, storefronts, and ambient lighting also make a path feel safer. If we could measure or estimate the average light level along each road segment, the model could become even more accurate.

In the long term, this project could be developed into a web or mobile app that helps people walking at night choose safer paths — starting with the UW–Madison campus, helping students move between buildings or bus stops while avoiding dark areas.

This project was originally completed as part of CS 524: Introduction to Optimization at the University of Wisconsin–Madison. It was a team effort, and I am now planning to continue this work as part of my ongoing research in real-world safety and smart routing systems.