SAFR | Devpost

App logo

We both were really passionate about doing something that would not only help broaden our horizons, like trying out something we have never done before, but also help our community in some way. Sophia walks to and from school, along with other places in our community, and she noticed that there are many instances of pedestrian safety hazards such as broken stoplights, unprecedented car accidents, lack of crossing guards, and many more. We came to the conclusion that this is not just a problem Sophia faced but something that affected everyone, whether they're a cyclist, a driver, or a pedestrian.

We were hoping to create an app that would allow community members to commute safely without fear, or in other words make safety accessible to everyone.

Our app provides safe route options and alternatives depending on the possible hazards nearby. It takes in safety concerns — such as lack of crossing guards, minor construction, and temporary lane reductions — from open source traffic data and categorizes them into two groups: high and lower risk concerns. We score possible routes based off of level of safety, and every hazard, depending on severity, reduces the score. Additionally, there is an AI insight explaining the risks of the route, which also specifies why it took points off.

We decided that this project would be too complicated to code right out, so we decided to go with vibe coding. To do this, we went to ChatGPT to give us a proper prompt, which we then pasted into v0, which helped generate an app/web page that would run the code we wanted it to run. We would constantly update the code, as well as any additional needed prompts, and continued to test the program over and over, making changes here and there, until it ultimately resulted in a program we approved of.

Throughout the developing process, we ran into many different problems including inefficient code, incorrect algorithms, and many more. However the biggest challenge we ran into was actually when we ran out of credits in v0, and went about the most complicated way to try and export the code to another account with the hopes of coding with that account, and then finally trying to upload the database of the original code onto that new account with the code. In addition to this, we also encountered a problem with the maps, originally the route that was generated was very, very, bad. However, we realized that we were not actually generating the route using any map, it was just drawing a random line. After finding out about this we changed the code to use the same map that is being displayed to generate the route.

We are proud of the fact that we persevered through the challenges we faced even when they proved to be very difficult to overcome. Although we may nto have been able to finish the program exactly as we originally intended, we were still able to get something out of it, and hope that later on, we would be able to develop the program to work better than originally intended.

In the future, we hope that SAFR would be used by civilians all around the world to help find safe, accessible, foolproof routes that allow them to commute safely.