rideOs TestKit

Calculations with Chicago dataset
Chrome extension (Austin dataset)

Inspiration

We were interested in finding the optimal way to send out autonomous vehicles based on ride share data. How much better would distributing parts of your fleet based off of forecasted demand be, than having the cars reactively assigned to ride requests? How could we best utilize un-used fleet vehicles?

Take the example of daily morning commute: if most fleet vehicles end up being sent to the city center during the morning, and the cars aren't actively re-distributed, would that drastically increase wait times for rides in the outskirts of the city?

What it does

Intakes ride share information from Austin, Texas
Uses rideOS to calculate trip wait times / travel times / routes
Compares two different methods of fleet vehicle distribution
Displays resulting metrics, routes in a Google chrome extension

How we built it

Used various data science packages within Python, Javascript and the Chrome store for the Google Extension, Leaflet, d3 and Mapbox for data visualization

Challenges we ran into

Some parts of the API weren't built to work together (e.g. Fleet Planner V2 isn't meant to be used . with the RideHail Driver API)
The API currently doesn't work for Chicago (which our original dataset was from)