Robot Learning 2026 – Lecture 1: Introduction to Robot Learning | ETH Zürich

Robot learning aims to bridge the gap between traditional, model-based control and modern, data-driven machine learning to create generalist robotic policies. While high-level reasoning is now easily achieved by AI, physical interaction remains challenging due to the scarcity of diverse robotic data and the complexity of real-world environments. This course introduces students to the fundamental algorithmic frameworks of imitation and reinforcement learning, moving beyond rigid, hand-coded state machines toward autonomous systems capable of adapting to unstructured tasks. By leveraging large-scale, multi-modal sequence modeling—similar to the paradigm shift seen in vision and language models—the field is approaching a "ChatGPT moment" for robotics. Students will gain hands-on experience with a fleet of SO101 robots, transitioning from individual simulation-based assignments to complex, real-world group projects that explore the future of physical intelligence.