Beacon AI

The public's concern for their safety is ever-increasing. People feel unsafe walking around their own homes at night. We endeavour to give them peace of mind and guide them using a safer path.

Comment

Inspiration

From personal experience travelling through unfamiliar areas as well as stories from those around us, we hope that by using our technical experience, we can make a difference.

What it does

Our product is a safe routing app that shows the user multiple paths each with a safety rating, allowing them to choose a balance between safety and speed when going from point A to B, with peace of mind that they aren't taking unnecessary risk.

How we built it

Our safety score is generated utilising an ML model built on a random forest linear regression; this is deployed using a flask backend that hooks into our react native frontend application.

Challenges we ran into

Our biggest challenges were the integration of both the frontend and backend, as well as being able to overcome the stress and fatigue from working together for such a prolonged time. Additionally, we had to pivot the idea quite late on; however, through productive discussion, we were able to settle on something that we were much happier with.

In the backend, training the model using an appropriate strategy to where the weights weren't nonsense was a task, however through research and trial and error we found a model applicable to our use case.

Accomplishments that we're proud of

We are proud of our teams perseverance through issues and transparency when someone felt uncomfortable with a part of the project; this lead to a much better end product that the entire team was much happier with.

What we learned

We learnt about integrating a stack and the challenges that come along with that, specifically when they are developed by two independent teams. Additionally, we learnt the value of challenging the status quo and tearing an idea apart as well as how important calm, composed leadership is.

What's next for Beacon AI

The next steps for Beacon AI are to continue the product, refining the aesthetics as well as improving the model over time as data is improved and expanded on. Additionally, the model can developed into a deep learning model that takes in more quantitative factors to become more accurate in its predictions. Ideally, this will also be released and hopefully start to make a difference as we all believe it can do.

Built With

Share this project:

Updates