Robot Learning

Lecture notes, slides, and recordings covering RL and deep learning for robotics.

Glen Berseth · Université de Montréal / Mila - Quebec AI Institute

Watch on YouTube Assignments WIP

This course covers the core algorithms and ideas at the intersection of deep learning and robotics, progressing from supervised imitation learning through model-based planning, policy gradient methods, value-based RL, goal-conditioned policies, reward learning, and practical deployment concerns. Slides and book chapters are drawn from three years of the graduate course at Université de Montréal.

Programming assignments for this course are available on GitHub (milarobotlearningcourse). Note that the assignments are a work in progress and will continue to be updated.

Lectures

Foundations

What Is Robot Learning?

Video Slides Chapter

Introduction to Deep RL

Behavior Cloning & Imitation

Supervised Learning & Behavior Cloning

Video Slides Colab Chapter

Data Collection for Robotics

Planning & Model-Based RL

Planning

Playlist Slides

Learning to Plan

Using Learned Models

Playlist Slides Colab

Policy Gradient Methods

Policy Gradients

Value Methods

Value-Based Methods

Q-Learning

Video Slides Colab

Sequence Models

Sequence Models & Auxiliary Learning

Goal-Conditioned & Language-Guided RL

Goal-Conditioned RL

Visual Goal-Conditioned RL

Language Goal-Conditioned RL

Reward Functions with LLMs

Reward Learning

Learning Reward Functions

Hierarchical RL

Hierarchical RL

Multi-Agent RL & HRI

Multi-Agent RL & Human-Robot Interaction

Offline RL

Offline RL

Transfer & Generalization

Sim-to-Real Transfer

Generalization for Robotics

Continual & Autonomous Learning

Lifelong Learning

Autonomous Learning

Changelog

2026-06-14 — Initial website launch with lecture slides, videos, and readings.