#4 Know-How Event: Spatial AI

From 3D Visual Perception to Embodied Foundation Models • WHAT TO EXPECT How do machines learn to see, understand and act in the physical world? That’s the question at the heart of Spatial AI, and the field is moving fast. At our #4 Know-How Event: Spatial AI: From 3D Visual Perception to Embodied Foundation Models, we look at how Spatial AI has grown from visual perception, 3D and 4D geometry and dynamic scene understanding towards vision-language models, world models, world action models and vision-language-action policies for physical agents. The talk connects both directions and examines their shared algorithmic challenges. It covers data-driven methods for 3D computer vision and spatio-temporal modelling, including monocular depth estimation, point cloud processing and trajectory forecasting. It then presents advances in modern Spatial AI with a focus on robotic arms and places them in the context of robustness. A compact deep dive straight from research for anyone curious where perception, robotics and foundation models are heading. SPEAKER Prof. Dr. Vasileios Belagiannis holds the Professorship for Machine Learning in Signal Processing at the Chair of Multimedia Communications and Signal Processing at Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). PARTICIPATION INFO The event takes place on site at ZOLLHOF (Zollhof 7, 90443 Nürnberg). All details will be sent to you by email in advance. Please check your Eventbrite account settings to make sure you can receive emails from Eventbrite. LANGUAGE Keynote will be held in English, Networking however can take place in English or German, whatever you prefer.