FareShare

Inspiration

-- tired of everyone asking to use your car? and worrying about if they will crash or damage it?

What it does

Keeps every driver of a car responsible for their insurance, car payments and the way they drive, depending on their usage

How we built it

authenticate vehicles to users using face recognition and password (encrypted biometrics)
track vehicle usage using OBD II data from CanBus, read with an OBD II scanner while the car is actively used
use a smart contract to orchestrate payments between the authorized users of the car
all authorized users make a deposit payment into the smart contract of the vehicle at the beginning of the month
the smart contract reads the vehicle usage for each user at the end of the month
the car (dealership) payment and insurance monthly premium are paid out from the smart contract at the end of the month
the smart contract refunds users based on how much usage they had (more usage, less refund) and the monthly cycle is complete
users can view their driving statistics and records at any time on the mobile app
users can also view payment records and history on the mobile app
we used the Algorand blockchain due to fast transaction speeds (<8 sec transaction time) as well as relatively low transaction fees

Hardware

The following clips demonstrate how we acquired readings from the OBD2 Scanner. https://drive.google.com/file/d/1r-Ejw1wWZkA38Bxl7kHzUsqg0x5Hcs_q/view?usp=drivesdk https://drive.google.com/file/d/1qwhGIroXAJkvbivTFMA7g9e9NH9KrHoe/view?usp=drivesdk https://drive.google.com/file/d/1qmPsk8gz1T1YRliIyZR4H1wOXTjhFUy_/view?usp=drivesdk https://drive.google.com/file/d/1qIpmyCZqBs3rdAd4Sa4uaukQtpoVYAwl/view?usp=drivesdk

Mantech Challenge

We used a 256 bit key to encrypt the face embeddings of the user, and this is further used as a seed to generate the mnemonic which allows access to the keypair of the Algorand wallet of the user.
https://www.youtube.com/watch?v=ECtsWnOZMxM
In this video, we show how this is done on the backend. The user is able to secure their face embedding biometric with a password or a mnemonic phrase, which can in turn be used to generate the wallet keypair for Algorand (uses a BIP39 mnemonic plus an extra word generated by checksumming the first 24 mnemonic words).

Peraton Challenge

We authenticate the user using face embedding based facial recognition. We obtain face embeddings from an image with a face by using a wrapper around the DLIB library (http://dlib.net/). The Dlib wrapper allows us to first locate faces within the image, and then extract a feature vector containing the face embedding. The embeddings are encrypted and stored with a password (or key) provided by the user. The following videos show this mechanism in action:
first user authenticating: https://youtu.be/IV38_FqHhmU
second user authenticating: https://youtu.be/UzTXB7FpV9M

Capital One Challenge

This project relies on making distributed payments using smart contracts on the Algorand Blockchain. We chose this Blockchain Platform due to low latency of transaction confirmations as well as very reasonable transaction fees. Each user has their own wallet, and the contract itself has a wallet, as do the insurance agency and the car dealership. The entire payments and transactions system is detailed in the following video:
https://www.youtube.com/watch?v=Ci1y14cd7jA