Pilot 01 · Vinhomes & Vincom Retail

Smart Buildings & Indoor Environmental Quality

Integrated occupancy, ventilation, and air-quality sensing across VinUniversity, Vinhomes residences, and Vincom Retail spaces.

Pilot Status: Planned | Formal Launch: January 2027 | Timeline: VinUniversity (2026) → Vinhomes Phase 1 (2027)

Executive Summary

The Challenge

Indoor environmental quality directly impacts occupant health, comfort, and productivity. Yet most sensor networks drift silently—CO₂ sensors age, calibrations fail, and occupants never know whether they're breathing truly fresh air or a sensor's false positive. Pilot 01 deploys self-aware sensing infrastructure that quantifies this uncertainty in real time.

Technical Architecture

Sensor & Inference Stack

Neuromorphic Edge-AI Inference

Environmental modules compute uncertainty metrics locally on-chip using low-power memristive hardware. This eliminates raw, un-vetted sensor streams being transmitted to cloud networks, enabling:

  • Zero-latency confidence scoring — confidence bounds computed before data leaves the node
  • Privacy by design — raw environmental streams never transmitted, only validated measurements and uncertainty estimates
  • Autonomous failure detection — sensor drift, bio-fouling, or thermal stress flagged on-node without cloud dependency

CMOS-Integrated Electrochemical Gas Sensing

Solid-state, fully integrated gas sensors (CO₂, TVOCs, particulate matter) engineered to execute automated on-chip baseline calibration. This directly addresses sensor drift:

  • Integrated AFE tuning — calibration baseline updated dynamically as the sensor ages
  • Cross-channel validation — multi-modal gas sensors detect electrochemical inconsistencies
  • Occupancy-adaptive baseline — CO₂ drift differentiated from occupancy changes via pattern recognition
Platform Specifications
Measurement Array: CO₂ (0–5000 ppm), VOCs (integrated), Particulates (PM2.5, PM10), Occupancy (PIR + CO₂ correlation)
Update Rate: Primary loop 1 Hz, uncertainty estimates 10 Hz, cloud sync 5 min
Power Budget: 2.5 mW (continuous), 1.2 mW (low-power mode)
Connectivity: LoRaWAN backbone with Zigbee fallback for in-building mesh
Live Demonstration

Adaptive Electrochemical Drift Compensation

Watch how on-chip RLS (Recursive Least Squares) filters detect and compensate for sensor drift in real-time. Adjust drift intensity and sensor type using the controls below. The algorithm continuously learns the baseline corruption pattern and corrects it before the signal leaves the node.

20%
0.03
Drift Magnitude
+0.0 ppm
Compensation Error
0.008
Efficiency %
98.5%
Matrix Condition
2.4
🔴 Raw Sensor (Corrupted) 🟢 Compensated Output (RLS Filter) ⚪ True Baseline (Reference) 📊 RLS Algorithm: Adaptive Least Squares Tracking
Rollout Timeline

Phase 1 & 2 Deployment

Phase 1 · VinUniversity (2026)

Controlled environment with known baseline. Academic buildings, dormitories, and common spaces instrumented with 40–60 sensor nodes. Quarterly validation cycles with ASHRAE comparators.

Timeline: Jan–Dec 2026
Phase 2 · Vinhomes Residences (2027)

Scale to residential occupancy patterns with seasonal variation. Initial deployment: 200+ units across one Vinhomes complex. Occupant feedback and long-term drift characterization.

Timeline: Q2 2027–Q4 2027
Core Research Frontiers

This pilot draws on V-SENSE's three core engineering frontiers:

Frontier 01
Self-Aware CMOS
Frontier 02
Neuromorphic Edge
Frontier 03
Reconfigurable RF