ViScan Smart Mirror

01

Problem

Camera, FMCW radar, ultrasound, and rPPG modules have different update rates and failure modes. Without explicit gates, stale frames, low light, multiple people, ambiguous radar targets, or calibration mismatch can propagate into the next measurement stage. ViScan therefore defines sensor responsibilities and state-transition conditions before combining outputs.

02

Architecture

The public system flow is Radar Presence → Face Session → AR Guide → Ultrasound Measurement → rPPG → Report → Session Reset. The BGT60TR13C supplies presence, distance, and stability information, while the camera handles face sessions, pose, exposure, occlusion, and multiple-person conditions.

03

My engineering contribution

• C# WPF and Python integration for the state-driven sensing pipeline
• YuNet/SFace face-session handling and MediaPipe Pose abdominal AR guidance
• Low-light, low-frame-rate, occlusion, multiple-person, and stale-frame checks
• Sensor fusion between BGT60TR13C presence/distance data and camera-side decisions
• Fail-safe transitions for timestamp, projection, and calibration mismatches

04

Public milestones

• CES 2027 Innovation Awards application submitted
• ViScan booth operation and live product demonstration planned for CES 2027
• Integrated test rates: radar 20.017 Hz, camera 14.696 fps, pose 14.650 Hz
• Technical reports, presentations, company meetings, prior-art review, and commercialization planning completed