Our team has concluded this week as the hardest, most confusing and most troubling week we had this year so far. After we took a step back to see where we had progressed to and after we analysed all the feedback we got from our mentors and our lectures we felt very unstable.
After many discussions and headaches we formalised and approved the following solution:
Need: To encourage usage in shared rides.
How: By improving trust in the drivers for the riders.
Implementation: Building an SDK for carpooling applications
For the matter of presentation:
**(and because we cannot act as an SDK on Waze, ideally this information will be provided by the service)
An application that confirms current carpooling via Bluetooth and notification confirmation.
After confirming a ride, the driver is registered in the history of the rider drivers.
Acquiring Data:
All users will be identified by their phone number, and give access to their list of contact whom they trust (possibly removing particular contacts whom they don't trust)
Each user will have a list of past rides
Using Data:
While searching for a new ride the user will see the potential riders with a *VERIFIED* marker or not if the driver appears in the past rides of one of his phone contacts.
The user has the option to see who previously shared a ride with the particular driver
Scalability:
Can expand to “ask about their experience”
Analyse driver photo using AWS Machine Learning algorithms
Backend notes:
Each user has a hash-table of past drives. A main-user has connections of contact-users. They also have a hash-table of past drivers.
Run BFS to find the new driver.