Robot Learning: A Tutorial
Francesco Capuano, Caroline Pascal, Adil Zouitine, Thomas Wolf, Michel Aractingi
2025-10-15
Summary
This paper provides an overview of the exciting changes happening in robot learning, highlighting how it's moving away from traditional programming and towards using machine learning techniques.
What's the problem?
Historically, robots were programmed with very specific instructions for every task, which was time-consuming and limited their ability to adapt to new situations. This meant robots struggled with anything outside of their pre-programmed routines and couldn't easily handle the complexities of the real world.
What's the solution?
The paper explains how recent advances in machine learning, particularly reinforcement learning and behavioral cloning, are allowing robots to learn from data instead of being explicitly programmed. It also discusses newer, more advanced models that can understand natural language and perform a variety of tasks on different robots, all using a practical toolkit called lerobot.
Why it matters?
This shift is important because it unlocks the potential for robots to be much more versatile and autonomous. Instead of needing a human to reprogram them for every new task, robots can learn and adapt on their own, opening up possibilities for robots to be used in a wider range of applications, like assisting in homes, exploring dangerous environments, and working alongside people in factories.
Abstract
Robot learning is at an inflection point, driven by rapid advancements in machine learning and the growing availability of large-scale robotics data. This shift from classical, model-based methods to data-driven, learning-based paradigms is unlocking unprecedented capabilities in autonomous systems. This tutorial navigates the landscape of modern robot learning, charting a course from the foundational principles of Reinforcement Learning and Behavioral Cloning to generalist, language-conditioned models capable of operating across diverse tasks and even robot embodiments. This work is intended as a guide for researchers and practitioners, and our goal is to equip the reader with the conceptual understanding and practical tools necessary to contribute to developments in robot learning, with ready-to-use examples implemented in lerobot.