Save Fuel

Inspiration

We could significantly save on fuel just by learning to drive a little smoother. Slower accelerations, wider turns, and keep that A/C down. Save Fuel provides real time feedback on your car's fuel consumption. By watching your consumption rates vary, you can learn to drive more efficiently!

What it does

Provides analysis of your average driving performance and your performance in relation to others. The predictive system utilizes a deep neural network to predict your fuel consumption in the near future, allowing you to accommodate and minimize your losses. We're playing with fate here!

How we built it

Save Fuel is built in Python. We used scientific computing libraries to handle large data files efficiently. The deep neural network is implemented in the open source library, Tensorflow.

Challenges we ran into

Predicting fuel consumption based on other (potentially unrelated) parameters proved to be a worthy optimization challenge. There were only a few trips in the data, and the sensors had very sporadic behaviour. Both problems required engineering to solve.

Accomplishments that we're proud of

Converging to a successful longer term average prediction was a huge accomplishment for us.

What we learned

The more data the merrier. Same goes for computational power. Other from that, did you know we can algorithmically determine what stores you stopped at on your trip?

What's next for Save Fuel

There are a lot of things that can be done with predictive analytics. Displaying them as feedback to the driver is just the beginning. We can envision a platform that could save the driver significant time and money using only the sensors provided on their car.