Abstract

Engineering laboratories contain electrical and electronic equipment that may be exposed to overheating, abnormal power consumption, unattended operation, unauthorized access, smoke or gas hazards, and other unsafe conditions. Conventional monitoring is often manual, reactive, and based on isolated sensor readings. This paper presents LabShield AI, a privacy-aware Artificial Intelligence of Things (AIoT) laboratory risk intelligence and decision-support architecture. The proposed system integrates temperature, humidity, power, occupancy, smoke/gas, laboratory access, and equipment-status information through an ESP32 edge controller. Sensor observations are buffered and transmitted through secure MQTT/TLS or HTTPS channels to a cloud/backend layer for validation, cleaning, normalization, and storage. Machine-learning components based on Random Forest or Decision Tree classification and Isolation Forest anomaly detection are proposed to analyze multi-parameter conditions. The system converts the analysis into a Laboratory Risk Score on a 0–100 scale with SAFE, WARNING, and CRITICAL states, and presents contributing factors and actionable recommendations through a web dashboard. Privacy-aware identifiers and role-based access control are incorporated to reduce unnecessary exposure of personal information. The architecture is motivated by recent research on AIoT laboratory monitoring, industrial predictive maintenance, IoT security, privacypreserving analytics, and IoT cybersecurity baselines [1]–[10]. The paper describes the system design, processing workflow, security model, and evaluation requirements. Since the supplied project material does not report measured experimental results, no performance values are claimed here; experimental validation is identified as necessary before publication as an empirical performance study.

Keywords

How to Cite This Article

Rights