Cyber-Physical Systems, Artificial Intelligence in Systems, Embedded Systems, Internet of Things (IoT)

Over the past three years, our Smart System Co‐Design (SSC) Lab has pioneered a suite of low‐cost, scalable sensing platforms and AI models to tackle diverse environmental challenges—from indoor air quality in schools to landslide detection in the Himalayas. This talk will weave together four key threads of our work:

Multi‐Pollutant Classroom Monitoring: An edge‐intelligent AQMS couples real‐time CO₂, VOC, PM, and gas‐sensor data with onboard, privacy‐preserving occupancy detection, revealing how human presence and HVAC strategies shape indoor air quality.

Forecasting with Timezone‐Aware LSTM & Hybrid Ensembles: We’ll showcase our novel timezone‐aware AR‑LSTM model and hybrid ensemble approach that accurately predict pollutant concentrations across multiple sites, improving decision support for ventilation control.

Drone‑based Remote Assessment: A custom, drone‐mounted PM₂.₅ sensing unit demonstrates cost‑effective, geospatial mapping of particulate pollution in remote regions, enabling rapid environmental assessment.

Transferable AI Frameworks for Hazards: Finally, we’ll highlight how Bi‑Directional LSTM networks and smart‐fire detection devices extend our methodologies to landslide prediction and precision agriculture.

Attendees will learn how tight integration of TinyML, edge analytics, and low‑cost hardware delivers actionable insights for air‐quality management, disaster prediction, and smart‐city applications—and how these architectures can be scaled to multi‑room, multi‑region deployments.