Back to Work

Motion Fusion Science Lab

A real-time science lab compares camera tracking with ultrasonic sensing and sensor fusion. I combined camera and ultrasonic inputs, then added saved experiments, graphs, exports, and reliability information.

Applied AI, computer vision, and full-stack engineerValidated prototypeAI / ML / Full Stack / Data
Motion Fusion Science Lab interface preview

What I built

The experimental system ingests camera and sensor data, calibrates the inputs, estimates motion, and preserves each trial for analysis.

  • Ingests browser camera and ultrasonic sensor data
  • Calibrates and fuses signals with uncertainty handling
  • Saves trials with charts, CSV export, and recommendations

My role

Applied AI, computer vision, and full-stack engineer

How it works

  • FastAPI and WebSocket services
  • React and Vite dashboard
  • OpenCV processing with SQLite experiment storage

Key decisions

  • Fuse complementary sensors with uncertainty weighting
  • Expose optional trained detection separately from the OpenCV fallback

What I tested

  • Calibration and anomaly checks
  • Saved experiment analysis
  • CSV export validation

Limitations and next steps

Experimental results are shown with confidence information and model limitations, not as final scientific conclusions.