1
SMART ROUTER
Team Members
INTRO
Gayathri
Daivik
Phani
Abhirut
Flaws in the current available routing systems
Flaws in the current system
Find an alternative route!
Around the corner...
What you should say
CONTRIBUTIONS
CONTRIBUTION
1. Road conditions – flooding, potholes – local drivers might already know such factors and avoid such routes
2. Type of traffic (Cyclists might want to avoid heavy vehicles)
Contribution
CONTRIBUTION
3. Crime (Avoid crime prone areas)
4. Garbage dumps (Cyclists or people on foot might want to avoid filthy neighborhoods)
Data source
DATA SOURCES
1. I change my city: Citizens post complaints pertaining to a wide variety of categories such as garbage, traffic, crimes, and potholes
2. Twitter : Crawled tweets mentioning potholes, Perform NER to recognize location, geocoded the location name to get the coordinates
-10 %
-40 %
-30 %
Data Source
Data Sources
3. Wikipedia: Extracted bus stop names, geocoded to get coordinates
4. Satellite imagery: Segmented satellite images to get location of green areas.
Satellite image segmentation
Satellite Image
Unsupervised as we have no labelled data
Used simple image processing techniques to extract a mask of the forest/ canopied areas
Images
Segmented Images
Our route avoids potholes!
DEMO
Other Quirks
1. Use a spacial hash to store and retrieve information points ( pothole/ crime etc) close to a given location
2. Precompute the edge lengths in the graph for every combination of preferences (like avoid garbage but do not favour greenery)