Researchers Develop Fully Autonomous Anomaly Detection System for Microcontrollers

Scientists deployed a TinyML anomaly detection system directly on low-power microcontrollers to monitor appliance behavior using power consumption data, eliminating the need for cloud computing or external connectivity. The STM32-based system achieved perfect detection accuracy with minimal processing latency and just 3.3 KB memory usage, proving robust AI is viable for resource-constrained embedded devices.