51.512 Reinforcement Learning for Embodied AI

Course description

This course introduces the principles and algorithms of Reinforcement Learning (RL) with a focus on their applications in embodied intelligence. This course focuses on agents that learn through interaction with physical or simulated environments. In this course, students will explore both classical and modern RL methods, from tabular approaches and function approximation to policy gradients and model-based techniques. You will also examine how these methods enable adaptive behaviour in robots and other autonomous systems. Through lectures, hands-on labs, and a project, participants will gain practical experience implementing RL algorithms, analysing their performance, and understanding their role in decision-making, planning, and human–AI collaboration.

Instructor

Matthieu De Mari, Malika Meghjani

Number of credits: 12