{"id":1211,"date":"2026-10-07T16:04:59","date_gmt":"2026-10-07T14:04:59","guid":{"rendered":"https:\/\/ai4society.luiss.it\/?post_type=ai4society_obs&p=1211"},"modified":"2026-10-21T11:54:54","modified_gmt":"2026-10-21T09:54:54","slug":"atlas-lab-adaptive-and-physical-ai-agents","status":"publish","type":"ai4society_obs","link":"https:\/\/ai4society.luiss.it\/en\/osservatori\/atlas-lab-adaptive-and-physical-ai-agents\/","title":{"rendered":"ATLAS LAB | Adaptive and Physical AI Agents"},"content":{"rendered":"\n
\n

ATLAS LAB – Adaptive and Physical AI Agents<\/h2>\n\n\n\n

The ATLAS Lab \u2013 Adaptive and Physical AI Agents<\/strong> is an advanced research laboratory dedicated to developing adaptive, agent-based, and physically situated Artificial Intelligence systems<\/strong>. The lab fosters interdisciplinary research that combines the fundamentals of machine learning with experimentation on robotic platforms, embedded systems, and distributed architectures. The goal is to design intelligent agents capable of continuous learning, adapting to dynamic contexts, and interacting effectively with complex physical environments.<\/p>\n\n\n\n

The ATLAS Lab\u2019s activities are structured around five main research strands. The first focuses on Agentic AI and multi-agent systems<\/strong>, involving the development of architectures for autonomous agents capable of planning, reasoning, and cooperating in complex environments, ranging from Large Language Model-based pipelines to decentralised systems. A second area focuses on Embodied AI and intelligent robotics<\/strong>, with applications ranging from ground robots and manipulators to drones and cyber-physical systems for industry, logistics, healthcare, and defence. The lab also develops Edge AI and decentralised intelligence solutions<\/strong>, designing efficient models for devices with limited computational resources through compression, quantisation, and federated learning techniques. Another research strand focuses on continual learning,<\/strong> with the aim of creating systems that can acquire new knowledge while retaining what they\u2019ve already learned. Finally, the lab develops agentic workflows and industrial applications based on the orchestration of specialised agents, retrieval-augmented generation techniques, and human-in-the-loop decision-making systems, thereby fostering the intelligent automation of business processes. <\/p>\n\n\n\n

By combining theoretical research with applied experimentation, the ATLAS Lab aims to contribute to the development of next-generation methodologies for Artificial Intelligence<\/strong> and to facilitate technology transfer for applications with a high economic and social impact. The laboratory collaborates with international research networks and participates in competitive European and national programmes, including Horizon Europe, ERC, MUR and PNRR, strengthening partnerships with industry in the fields of Industry 4.0, autonomous logistics, robotics for healthcare, and intelligent systems.<\/p>\n\n\n\n

In line with European and national strategic priorities in the field of Artificial Intelligence, the ATLAS Lab helps strengthen Luiss\u2019s international scientific standing<\/strong> by fostering the development of highly specialised skills and encouraging the attraction of competitive funding and collaborations with public and private partners. Finally, the laboratory aims to serve as a reference point for research and innovation in the field of adaptive and agent-based AI.<\/p>\n<\/div>\n\n\n\n

\n
\n

Coordinator<\/h2>\n\n\n\n