Real-Time Traffic-Sensor Fusion for a Smart Intersection
Overview
What this challenge is about.
Design a fusion architecture: per-modality timestamping (NTP-disciplined within 50ms), per-modality confidence weighting (modality-specific noise model), Kalman-filter or particle-filter fusion at 1Hz, output: occupancy + confidence per lane per intersection. Implement reference pipeline in Python or Rust running on an edge box (NVIDIA Jetson Orin Nano at the intersection cabinet) with MQTT publish to central. Evaluate against frame-by-frame ground-truth video for 1 week at 3 pilot intersections — measure: per-modality vs fused MAE, disagreement reduction, latency. Deliver: architecture spec (10 pages), reference pipeline + tests, evaluation report (6 pages), 4-page rollout plan for 12 pilot intersections.
The Brief
What you'll do, and what you'll demonstrate.
Design a real-time sensor-fusion architecture that cuts cross-modality disagreement from 12-15 percent to under 4 percent at 12 pilot intersections.
Earning criteria — what you'll demonstrate
- Design multi-modality sensor-fusion architectures with explicit noise models
- Implement Kalman or particle filters for real-time cyber-physical systems
- Validate fusion outputs against ground-truth video
- Specify edge-compute + MQTT architecture for municipal-grade reliability
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Career mappings coming soon.