Explore the intersection of robotics and imitation learning in this 54-minute talk by Matteo Saveriano from the Alan Turing Institute. Delve into the challenges of programming robots to operate effectively in human environments, focusing on precise movement planning and physical interactions. Discover how imitation learning techniques can teach robots new skills without explicit coding, extracting patterns from minimal demonstrations and applying them to various scenarios. Examine the limitations of traditional Euclidean-based approaches and learn about an innovative method that incorporates differential geometry to handle non-Euclidean data common in robotics. Gain insights into recent experimental results and current open issues in the field of geometry-aware imitation learning for robotics.