Software and Hardware for AI

Our department is also active in the field of artificial intelligence, with applications in: robotics and nonlinear control, neuroscience, big data man-agement and analysis, energy transition and sustainability, agentic AI, and knowledge representation, also addressing challenges related to the deployment of AI models in distributed environments. In robotics and nonlinear control, research topics range from intelligent perception and control for robots; through AI-based methods for fundamental nonlinear optimal control, networked systems, and estimation; to fundamental algorithms and software for reinforcement learning and deep neural networks. These methods are applied in marine, ground, and aerial robotics, precision agriculture, medicine, rehabilitation robotics, and more. Research in neuroscience carried out by the department’s team involves developing machine learning techniques for spike sorting and burst detection, EEG data analysis, and mapping functional brain networks using graph neural models. In collaboration with the Transylvanian Institute of Neuroscience, the team aims to develop computational tools for analyzing neural and brain activity. In the area of big data, the teams develop and adapt advanced AI techniques, including Transformer and LLM models, deep neural networks, large-scale language models, and reinforcement learning, to support decision-making and enable predictive as well as descriptive analytics (e.g., predictive maintenance, modeling user behaviors in household scenarios, energy consumption profiling, etc.). Research also explores scalable storage systems, visualization techniques, and preprocessing methods for data cleaning and transformation. While the main focus is on IoT data, the developed techniques are not limited to it and can be applied to other types of data. In the direction of energy transition and sustainability, research topics include advanced AI, blockchain, and distributed control techniques to support decision-making in energy network management and to perform predictive analyses related to energy demand and supply, consumption optimization, and renewable integration. These approaches enable the development of decentralized, secure, and privacy-preserving energy markets as well as the modeling and digitization of industrial processes. They also include the design and implementation of smart contracts for automated energy transactions, AI mechanisms for peer-to-peer trading without the need for a trusted intermediary, and tokenized ecosystems to incentivize distributed production and prosumer participation Research themes also include the development of explainable AI models, by creating algorithms capable of explaining decisions—a crucial area in the application of AI in sensitive sectors. Agent-based technologies target: