WaspadaOjol

Stay Alert, Stay Alive, with the AI-Powered Pedestrian Safety app. WaspadaOjol, inspired by the daily amount of people that walk around the streets without a care of their surrounding.

Comment

Inspiration

Every day, pedestrians walk around with their headphones on or their heads on the phone. This made us realize, people tend to be careless about the smallest things that could cause the biggest problems. Because of that, WaspadaOjol was created. The name was inspired by the popularity and amount of usage of motorcycles in Indonesia, and the fact that motorcycles are speedy vehicles. But it doesn't mean it is only applicable to motorcycles only, other vehicles are too.

WaspadaOjol was created to make pedestrian safety more accessible just by their phone, using AI to listen, analyze, and provide instant warnings when danger is detected.

What it does

WaspadaOjol is an AI-powered road safety assistant that detects approaching vehicle sounds and alerts users in real time.

The app:

  • Uses AI audio recognition to identify potential vehicle threats
  • Provides instant visual warnings through an emergency warning screen
  • Plays warning sounds and sends notifications
  • Uses haptic feedback to alert users through vibration
  • Records safety sessions and tracks detected warnings

By turning sound into actionable safety alerts, WaspadaOjol helps users stay more aware of their surroundings.

How we built it

WaspadaOjol was built using Flutter for the mobile application and AI-based audio processing for threat detection. Originally, WaspadaOjol started as a WPA (Web Progressive Application), but then attempted an upgrade to become a real app.

Technology used:

  • Flutter/Dart for the cross-platform mobile application
  • YAMNet AI model for audio classification and sound recognition
  • TensorFlow Lite for running AI locally on the device
  • Microphone audio streaming for real-time sound input
  • Local storage for saving safety session data
  • Notification and vibration systems for multi-sensory alerts

The system continuously analyzes environmental audio and triggers a warning when a dangerous sound pattern is detected.

Challenges we ran into

One of our biggest challenges was being able to make a real-time audio identifier, with the given time limit, that allowed the app to give the warning screen during the street mode.

Other challenges included:

  • Processing real-time microphone data efficiently
  • Creating fast alerts without causing unnecessary warnings (although Yamnet we tested it beforehand)

Accomplishments that we're proud of

Finishing alone and got to make an app for this competition.

What we learned

Through building WaspadaOjol, we learned that creating impactful technology requires more than just building a model, it requires understanding users, designing reliable systems, and considering real-world limitations.

We learned about:

  • Deploying AI models on mobile devices
  • Processing real-time sensor data
  • Designing user-focused safety experiences
  • Improving AI reliability through testing and iteration

What's next for WaspadaOjol

In the future, WaspadaOjol hopes to be able to tell the user in which direction the danger is coming from, and upgrade it's capability to real-time audio identifying.

Built With

Share this project:

Updates