Dimitris Chrysostomou

Embodied AI: Shaping Tomorrow's Robotic Intelligence

As we enter the era of generative AI, like ChatGPT, the potential of AI seems limitless. Yet, the future of robotic AI lies in embodiment - the seamless integration of perception, learning, and action in physical systems. This talk explores the frontier of embodied AI in robotics, examining how it surpasses traditional AI by adapting to our unstructured and uncertain world. We'll discuss recent breakthroughs, ongoing challenges, and the potential for creating intuitive, adaptive robots that can reliably operate in dynamic real-world environments, without relying solely on massive datasets or complex architectures. By bridging the gap between digital intelligence and physical embodiment, we want to shapea future where robots can truly understand and interact with the world as we do.

Bio: As a robotics and AI researcher at Aalborg University, I focus on developing reliable robotic solutions for safe human-robot interaction. With over two decadesof experience in EU-funded projects and numerous publications, my main research focus is to advance Intelligent and Collaborative Robotics for real-world applications, with a particular interest in robot behavior in industrial and social contexts, and the integration of AI ethics.