Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Deemphasising Dead-ends: Navigation in Today's Dendritic Cities


Published on

NACIS 2016 Presentation
Nate Wessel, University of Toronto
Algorithmic detection of dead-ends and highly indirect streets could help cartographers reduce visual noise in transport maps, without resort to generalization techniques that simplify data or remove it entirely. In this presentation, I'll discuss algorithms for detecting dead-ends and apply them to a sample of regions, using OpenStreetMap data. I'll attempt to show how the resulting classification can be used to reduce visual noise and make maps easier for the eye to navigate. Preliminary results show that dead-ends make up between 12% and 45% of all streets and/or paths in my broad sample of regions, and can depend in varying degrees on the chosen transport mode for which the network is constructed. The proposed technique then has special relevance for mode-specific transport maps or maps for users with unique access constraints.

Published in: Design
  • Be the first to comment

  • Be the first to like this

Deemphasising Dead-ends: Navigation in Today's Dendritic Cities

  1. 1. Deemphasizing Dead-Ends Navigating Today’s Dendritic Cities Nate Wessel PhD student at the University of Toronto
  2. 2. Background Dead-ends are frustrating. That’s why they put signs up. ->
  3. 3. What’s going on here? Map #1 - All streets are the same Map #2 - Dead-ends are 50% transparent - Indirect streets are up to 50% transparent 1 2
  4. 4. What’s going on here? Developed technique for a bike map in a hilly suburban city -> Problem: - Noisy map in the absence of road hierarchy Solution - Clarify topology - Reduce noise - Free up visual channels
  5. 5. So what are dead-ends, anyway?
  6. 6. So what are dead-ends, anyway?
  7. 7. Variability by mode
  8. 8. Variability by mode
  9. 9. Research Questions 1. How many dead-ends are there? 2. Where are they? 3. How dependant are they on mode? 4. Is this even remotely helpful??
  10. 10. Methods 23 urban regions selected for variety and data quality (whole built-up region used) OpenStreetMap -> osm2po -> PostGIS -> R Constructed graphs for: car, bike, foot Code on Github at
  11. 11. Results: How many dead-ends are there?
  12. 12. Results: How many dead-ends are there? Sorted by average
  13. 13. Results: How many dead-ends are there? Sorted by variance
  14. 14. Results: Where are they? ● Edge centroids -> KDE surface weighted by edge length ● Ratio of dead-ending segments to total ● = % dead-ends, locally
  15. 15. Results: Where are they? (Cincinnati by car)
  16. 16. Results: Dependence on mode?
  17. 17. Results: Dependence on mode? Anecdotal musings: - European cities and New Urbanism - Dead-ends for cars, connections for people
  18. 18. Next Step: Does this actually work? Need to test whether this technique actually helps people read maps ● Developing platform to test ○ A->B route-finding speed ○ A->B route-finding accuracy ○ Does it matter where A & B are? What if they are on dead-ends?
  19. 19. Discussion - Technique seems relevant to many cities - Mode specificity is an issue - Data quality issues a concern ( OSM ) - Algorithm could be way smarter - Needs empirical testing
  20. 20. The End (for now) Nate Wessel University of Toronto