Inspiration

Emergency vehicles lose critical time at traffic lights because city signals are usually static and disconnected. We built AmbuGUARD to explore a simple but high-impact question: what if traffic infrastructure could think, react, and prioritize life-saving movement automatically?

What it does

AmbuGUARD tracks an ambulance route, predicts upcoming signal conflicts, and coordinates traffic lights across intersections to help the ambulance pass with minimal delay. Instead of treating each junction separately, the system acts as a central agent that manages the corridor as one connected flow.

How we built it

We built AmbuGUARD as a simulation-first demo with a real-time control mindset. The system uses a map-based route flow, an agent-style decision layer, and live visual signal updates to show how emergency priority could work in a city. For a Microsoft-aligned version, the agent layer can be presented as built on Foundry Agent Service or Semantic Kernel, with custom tools handling route analysis and signal commands.

Challenges we ran into

The hardest part was designing believable coordination across multiple intersections while keeping the demo simple enough for a hackathon. We also had to balance realism, responsiveness, and clarity so judges could immediately understand the value of the system.

Accomplishments we’re proud of

We created a working concept that turns a real civic problem into an intelligent agentic workflow. The demo is easy to understand, visually strong, and scalable to real-world emergency mobility use cases.

What we learned

We learned that traffic optimization is not just about automation; it is about decision-making, timing, and orchestration. We also learned how to present a complex infrastructure idea in a way that feels practical and judge-friendly.

What’s next for AmbuGUARD

Next, we want to connect the system to live GPS inputs, add predictive traffic intelligence, support multiple ambulances and edge cases, and evolve the agent into a deployable smart-city control layer.

Share this project:

Updates