Better Buzzes

• 

  The app displays the buses according to closest stops.

• 

  Display of the bus route, along with the expected volume.

• 

  The app has better communication of bus statistics for users to access, such as the final stop that is reached.

• 

  The planned schedules of the bus routes to let users make future bus riding plans.

• 

  Allows for users to schedule the buses they will board, making them receive notifications.

Inspiration

That feeling where you are waiting for a bus and desperately have somewhere to be, we know it's frustrating! Current bus routes often have unreliable time estimations as well as counterintuitive user interfaces. College students are in need of a new bus system that prioritizes their movements around campus and notifies them of recent changes in schedule. This is why we created Better Buzzes, a mobile app that utilizes reinforcement machine learning to create optimal bus schedules that reduce waiting time and provide a better riding experience.

What it does

• Allows users to wait in virtual lines at each bus stop.
• Compiles user arrival times and wait times.
• Creates a model of the overall student body's weekly schedule.
• Utilizes reinforcement machine learning to create optimal bus schedules that reduce waiting time.
• Includes a notification system that notifies the user of future changes in schedule.
• If implemented on a larger scale it can help reduce transportation waiting time in supply chain processes.

How we built it

To build the app we first created a model that would be able to simulate the GT bus system given already known mass student behavior. Then, utilizing a policy gradient method, the buses schedules were altered over many iterations to optimize the total student wait time.

Challenges we ran into

• Learning UI development with Swift UI without any prior experience
• Distinguishing the key features needed to create an easy to use interface
• Conceptualizing the idea while maintaining a minimalist approach

Accomplishments that we're proud of

• Created new upgraded UI for the GT bus system
• Committing to our own project even though there were other challenges we wanted to participate in
• We helped improve GT!

What we learned

• A positive mindset can go a LONG way!
• Putting a function app together is much harder than it seems.
• Our best strategy was dividing/assigning all of our tasks from the beginning.

What's next for Better Buzzes

• Implementing and analyzing the data here at Georgia Tech!
• Upon those findings improve to make our process applicable in other scenarios.

Built With

• google-cloud
• python
• swift