Robot Learning Course

Politecnico di Torino · A.Y. 2026/2027

Physical AI: reinforcement & imitation learning for embodied agents that perceive, learn, and act in the real world.

Explore the Course

Photo: Michael Martinelli on Unsplash

Course Description

This course will provide foundational knowledge on Physical AI and high-level control, aimed at developing embodied agents able to perceive, learn, and act in the real world. Lectures focus on machine and deep learning solutions for autonomous systems, especially Reinforcement and Imitation Learning techniques behind today's generalist robot policies.

  • Physical AI
  • Reinforcement Learning
  • Imitation Learning
  • Sim-to-Real
  • Vision-Language-Action
  • World Models

Lectures

Hands-on labs

Papers discussion on recent research

Link to this page

Teachers

Giuseppe Averta
Giuseppe Averta

Associate Professor at Politecnico di Torino

giuseppe.averta@polito.it

Office @ Corso Ferrucci 112, Vandal Lab

Francesca Pistilli
Francesca Pistilli

Assistant Professor at Politecnico di Torino

francesca.pistilli@polito.it

Office @ Corso Ferrucci 112, Vandal Lab

Andrea Protopapa
Andrea Protopapa

PhD Candidate at Politecnico di Torino

andrea.protopapa@polito.it

Office @ Corso Ferrucci 112, Vandal Lab

Davide Buoso
Davide Buoso

PhD Candidate at Politecnico di Torino

davide.buoso@polito.it

Office @ Corso Ferrucci 112, Vandal Lab

Weekly Course Schedule

Day Activity Time
Monday Lecture (room 1I) 16:00 - 19:00
Friday Lecture (room 7T) 16:00 - 17:30
TBD Laboratory Session (room TBD) TBD

Course Modules

From the fundamentals of robot sensing and control to cutting-edge policy learning and Physical AI foundation models.

  • Foundations of sensing and actuation
  • Kinematics and dynamics of autonomous systems
  • Motor control
  • Introduction to ROS programming (optional)
  • Gaussian Filters, Kalman Filter, Particle Filter (optional)

  • Supervised Learning for Robotics
  • Imitation Learning vs Reinforcement Learning
  • Markov Decision Processes

  • Planning by Dynamic Programming
  • Model-Free Prediction
  • Model-Free Control
  • Value Function Approximation

  • Finite Difference Policy Gradient
  • Monte-Carlo Policy Gradient
  • Actor-Critic approaches
  • Natural gradient
  • Deep policy networks

  • Model-based learning
  • Integrated Architectures
  • Learning and Planning
  • Real and simulated experience
  • Monte-Carlo Tree Search
  • Go use-case

  • Intro to bandits
  • Simple naΓ―ve exploration (πœ–-greedy)
  • Optimistic approaches
  • Probability matching & Information Value
  • Multi-armed Bandits
  • Contextual Bandits

  • Behavior cloning
  • Imitation Learning (DAgger)
  • Generative IL (Diffusion Policies)
  • Inverse RL (Max-Entropy IRL, AIRL)
  • GAIL (Generative Adversarial Imitation Learning)

  • Sim-to-Real Transfer
  • Vision and Graph-Based RL
  • World Models
  • Vision-Language-Action Models

  • Optimal Control
  • Reward Shaping
  • Q-Learning (Value-Based RL)
  • Reinforce, PPO, SAC (Policy Gradients)
  • Sim-to-Real Policy Transfer

Resources & Textbooks

Everything you need to go deeper, from foundational textbooks to the frameworks we use in the labs.

Reinforcement Learning: An Introduction

Sutton & Barto
Deep Learning

Goodfellow, Bengio & Courville
Robotics Modelling, Planning and Control

Siciliano et al.
Probabilistic Robotics

Thrun et al.
Gymnasium

RL environments collection
PyTorch

Deep learning framework
Stable Baselines3

RL algorithms in PyTorch
OpenAI Spinning Up

RL educational resource
David Silver's RL Course

DeepMind lectures & notes
imitation

Behavior Cloning, DAgger, GAIL, AIRL in PyTorch
Diffusion Policy

Visuomotor policy learning via action diffusion
LeRobot

Hugging Face library for real-world robot policies & VLA models
DreamerV3

Mastering diverse domains through world models
World Models

Ha & Schmidhuber, interactive paper

Weekly Lecture Schedule & Materials

Links for slides will work only if you are logged in with your PoliTo credentials. You can find lecture recordings on the university portal approx. 1 day after the lecture.

Week Day Time Topic Covered Slides Material
1 Mon 21/09/26 16:00 - 19:00 L0: A few info
L1: Introduction
1 Fri 25/09/26 16:00 - 17:30 L2a: Kinematics of Robot Manipulators - pt1
2 Mon 28/09/26 16:00 - 19:00 L2a: Kinematics of Robot Manipulators - pt2
L2b: Dynamics of Robot Manipulators
L2c: Mobile Robots
2 Fri 02/10/26 16:00 - 17:30 L3: Motor Control
3 Mon 05/10/26 16:00 - 19:00 L4a: Intro to Machine Learning and Deep Learning - pt1
3 Fri 09/10/26 16:00 - 17:30 L4a: Intro to Machine Learning and Deep Learning - pt2
4 Mon 12/10/26 16:00 - 19:00 L4b: Intro to Reinforcement Learning
L5a: Markov Decision Processes - pt1
4 Fri 16/10/26 16:00 - 17:30 L5a: Markov Decision Processes - pt2
5 Mon 19/10/26 16:00 - 19:00 L5b: Dynamic Programming
5 Fri 23/10/26 16:00 - 17:30 L5c: Model-Free Reinforcement Learning – Prediction and Control - pt1
6 Mon 26/10/26 16:00 - 19:00 L5c: Model-Free Reinforcement Learning – Prediction and Control - pt2
L5d: Model-Free Reinforcement Learning – Value Function Approximation - pt1
6 Fri 30/10/26 16:00 - 17:30 L5d: Model-Free Reinforcement Learning – Value Function Approximation - pt2

Past Editions

Browse previous editions of the course, frozen as they were taught.

A.Y. 2025/2026