BusLive

Real-time bus tracking powered by AI. Get live locations, accurate ETAs, & smart traffic insights for a stress-free commute.

Comment

💡 Inspiration

I've been there—standing at a bus stop, checking my watch, wondering if the bus has already left or if it's just stuck in traffic. The uncertainty of public transport doesn't just waste time; it creates anxiety and disrupts daily schedules.

For students and daily commuters, this "blind waiting" is a daily struggle. I realized that while ride-sharing apps like Uber have democratized real-time tracking for private rides, public transport (especially in developing regions or campuses) remains largely offline and opaque.

I wanted to bridge this gap. I asked myself: Why can't catching a bus be as predictable as ordering a pizza?

🚌 What it does

BusLive is a comprehensive, real-time bus tracking ecosystem that brings transparency to public transit.

  • Real-Time Live Tracking: Users can see the exact location of their bus moving on the map in real-time, not just static updates.
  • AI-Powered Insights: I integrated Google Gemini AI to analyze traffic conditions. Instead of just showing a red line on a map, the AI explains why there is a delay (e.g., "Heavy congestion reaching the campus gate") and offers recommendations.
  • Smart Alerts: The app creates a geofence around the user. You get a notification when the bus is 5 minutes away, so you never miss a ride.
  • Dynamic ETAs: I calculate accurate arrival times based on distance and average speed, updating instantly as the bus moves.

⚙️ How I built it

I built BusLive with a mobile-first mindset, knowing most users would be tracking on their phones.

  • Frontend: I used Next.js 16 with React 19 for a blazing-fast, server-rendered interface. Tailwind CSS and shadcn/ui helped me create a premium, accessible design that looks great in Dark Mode.
  • Maps & Routing: I utilized Leaflet.js for interactive maps and OpenRouteService to snap bus coordinates to actual roads, ensuring the bus doesn't look like it's flying over buildings.
  • Real-Time Engine: Firebase Realtime Database was the backbone for syncing coordinates. It allows sub-second latency updates from the bus hardware to the user's screen.
  • The Brain (AI): I used Google's Gemini API via a custom Genkit flow. I feed it the bus's speed, location context, and time of day, and it returns a human-readable analysis of the commute conditions.
  • Hardware Simulation: I built a robust simulation engine that mimics realistic bus movement, including varying speeds, stops, and traffic delays, to demonstrate the platform's capabilities without needing a physical bus fleet.

🚧 Challenges I ran into

  • The "Jumping" Bus Problem: Raw GPS data is often noisy. When I first plotted points, the bus would "teleport" or jitter on the map. I had to implement interpolation logic to make the movement smooth and realistic between update intervals.
  • Context-Aware AI: Initially, the AI would just say "Traffic is bad." I had to fine-tune the prompts and provide structured inputs (speed variance, historic data) to get helpful insights like "Moderate traffic, but moving well. Expect a 2-minute delay."
  • Geofencing Logic: Calculating distances is easy; knowing when to trigger an alert without spamming the user was hard. I had to implement state management to ensure the "Bus Approaching" alert only fires once per meaningful event.

🏆 Accomplishments that I'm proud of

  • Butter-Smooth UI: The map experience feels native. The car icons rotate with the bearing of the road, and the transition between dark and light modes is seamless.
  • Live AI Integration: It’s not just a chatbot; the AI is an active observer of the trip, pushing insights automatically.
  • Scalability: The architecture is decoupled. I could plug in a real hardware tracker (ESP32) tomorrow, and it would work without changing a single line of frontend code.

🧠 What I learned

  • Maps are hard: Coordinate systems, different projections, and calculating accurate bearings took a lot of trial and error.
  • Real-time is a mindset: You can't just fetch data once. You have to design for streams, subscriptions, and connection drops.
  • AI adds empathy: Adding the AI layer transformed the app from a cold utility tool into a helpful assistant that "understands" the commute.

🚀 What's next for BusLive

  • IoT Integration: deploying actual ESP32 GPS modules on university buses for a pilot run.
  • Crowdsourcing: Allowing users to report on-ground incidents (accidents, breakdowns) to feed back into the AI model.
  • Predictive Scheduling: Using historical data to generate "true" bus schedules that account for daily traffic patterns automatically.

Built With

Share this project:

Updates