Driving Energy Intelligence through Embedded AI

As part of the IPCEI ME/CT program, co-financed by the Czech National Recovery Plan through the Ministry of Industry and Trade Czech Republic, we develop self-learning algorithms that run directly within devices, turning grid data into clear, actionable insights. This allows customers to make faster, informed decisions without cloud dependency.

SLAID (Self-learning Autonomous AIoT Devices) represents an innovative approach to embedded machine learning in the energy sector. The key focus of the project is to develop autonomous machine learning modules specifically designed for integration into smart grid devices with limited computational resources – such as smart meters, PV inverters, and grid controllers that form the backbone of modern energy distribution. Unlike traditional AI/ML solutions, which require significant computing power and cannot adapt to local conditions, SLAID enables intelligent processing directly within devices operating under strict power and memory constraints.

SLAID delivers autonomous edge intelligence modules powered by proprietary mBORT algorithms (multi-target Boosted Oblivious Regression Trees). These modules:

• learn and adapt automatically within the device,
• operate on low-power microcontrollers,
• don’t require cloud connectivity for core functions,
• provide real-time predictions, anomaly detection, and more.

Unlike conventional approaches, our algorithms learn continuously from small datasets directly within the device, eliminating the need for extensive historical data storage. We call this approach “Incremental Learning”.

Our architecture supports multiple chip types, including ARM, RISC-V, and FPGA, primarily from European semiconductor partners, aligning with Europe's technological sovereignty goals.

From predicting of household energy consumption to photovoltaic power plant’s production, our modules enable smarter energy management at the grid edge.