Physical Intelligence

Fall 2025. ESE 6510. Tue / Thu 10:15-11:45. Fagin Hall 118.

Image

Announcements

First day of class.

Aug 26 · 0 min read

I will be giving a talk at the Meta workshop on egocentric perception on Aug 25-26. The first day of class will be after the workshop.

Course Overview

Why don’t we yet have a foundation model that can operate robots in the physical world? What makes applying the standard pre-training approach so uniquely challenging in robotics? And what recent advances are pushing the boundaries of what’s possible in real-world applications?

This course offers a structured framework to explore these questions. We will study the mathematical foundations of techniques for learning-based decision-making and examine how these methods are deployed in robotic systems. Through a combination of lectures, hands-on projects (including a drone race!), and guest presentations from leading experts in the field, students will gain a deep understanding of the state-of-the-art decision-making systems and their challenges when applied to robotics.

Prerequisites

This is a graduate-level course. Students are expected to have prior knowledge in deep learning and robotics, such as the topics covered in Robot Learning (ESE 650), Applied Machine Learning (CIS 5190), and Introduction to Robotics (MEAM-520).

Schedule

Foundation

AUG 28
Intro
Syllabus,
Slides
SEP 2
What is a Robot? Actuation
Slides
SEP 4
What is a Robot? Perception
Slides
SEP 9
Intro to Probability I
Notes (class content was slightly different)

Reinforcement Learning

SEP 11
Intro to Reinforcement Learning
Slides
SEP 16
Intro to Reinforcement Learning
Slides
SEP 18
Policy Gradient
Shulman, Ch. 2.6
SEP 23
Policy Gradient (Advantage Estimation)
Shulman, Ch. 4
SEP 25
Tutorial on RL
Slides
SEP 30
Advanced Policy Gradient (TRPO,PPO)
Shulman, Ch. 3, PPO
OCT 2
Policy Gradients and World Models
Slides
OCT 7
Value and Q Learning
S&B, Ch. 4; Ch. 5(up to 5.5); Ch. 6(6.5,6.7)

Sim2Real

OCT 9
No Class Fall Break :fallen_leaf:
OCT 14
Building a Simulator
Slides
OCT 16
Sim2Real: Intro & Basics
Slides, Race challenge is out
OCT 21
Sim2Real: Tips & Tricks and Intro to Imitation Learning
Slides

Imitation Learning

OCT 23
Learning from Pre-Trained Models (Behavioral Models)
Slides
OCT 28
Imitation Learning as Generative Modeling: Flow-matching, Diffusion
Slides
OCT 30
MidTerm

Applications and Frontiers

NOV 4
Guest Lecture I
Homanga Bharadhwaj
NOV 6
Guest Lecture II
Rachel Holladay
NOV 11
Tutorial on Imitation Learning
Slides
NOV 13
Guest Lecture III
Yunzhu Li
NOV 18
Guest Lecture IV
Haozhi Qi, Submission Deadline for Race Phase I.
NOV 20
Challenges ahead
Slides
NOV 25
The Illusion of Intelligence
Slides
NOV 27
No Class Thanksgiving :turkey:
DEC 2
Recitation
DEC 4
Race Day (Phase II)

Instructors

Avatar

Robots that learn, UC Berkeley.

Robotics Manipulation, MIT.

Deep Reinforcement Learning: CMU version UC Berkeley version.

Introduction to Robot Learning, CMU.

Embodied AI Safety, CMU. This course is not only very interesting, but also has an awesome webpage.