Inspiration
Car insurance is too general. While the industry makes efforts to customise risk profiles they currently do not take into account driving frequency, driving routes or driving habits. Our innovation allows drivers to record their driving data and leverage it lower insurance premiums. Importantly drivers maintain power over their data and can make it available via the IOTA tangle to any service prvoiders. As a proof of we have built risk models based on VicRoads and road typography to calculate risk profiles for individual drivers.
What it does
Our solution combines, IOT, Machine Learning, Blockchain and Serverless Architecture to enable a Pay as You Go insurance model that allows users to be offered customised insurance plans based on how they actually use their car.
We do this by building a cloud connected GPS device that sends a users GPS data to the cloud where we use ML to process the points and calculate a risk score. This risk score as well as advice on how to lower your risk is provided in the front end combined with a map showing your journey and risks you took travelling on different roads.
On top of this we send our data to the IOTA Tangle (Blockchain like architecture) where users will be given a unique key so they can share their data with any car insurance company they like. This puts the power of the users data back into the hands of the user and allows them to control what happens to their data.
Challenges I ran into
IOTA Tangle is incredibly difficult to work with (although useful) due to lack of documentation. Combing the IOT, ML, IOTA and Serverless Architecture into one working application was painful.
What's next for trackr
Using the wealth of collected GPS data to build new applications for transport within Australia
Frontend: http://trackr.datath.com/index.html?location=melbourne
Link to github repo:
https://github.com/letianw91/unihack18_ML
https://github.com/letianw91/unihack18_backend
https://github.com/letianw91/unihack18_frontend
https://github.com/letianw91/unihack18_IOTA
Log in or sign up for Devpost to join the conversation.