Overview
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Learn about Temporal Difference Model Predictive Control (TD-MPC) through a detailed technical video that breaks down the problem formulation and rollout mechanics. Starting with notation and problem formulation fundamentals, explore the high-level concepts of Model Predictive Control (MPC) before diving into fixed horizon optimization. Examine the Cross-Entropy Method (CEM) through both theoretical frameworks and practical physics-based thought experiments, culminating in its application to action trajectories. Gain insights from this first installment of a two-part series that references the original TD-MPC research paper and includes comprehensive explanations of key concepts across multiple chapters.
Syllabus
- Intro
- Notation and problem formulation
- High level summary of MPC
- Why are we optimizing for a fixed horizon?
- Generalizing to a formulation for CEM
- CEM with a physics thought experiment
- CEM applied to action trajectories
- Summary
Taught by
Hugging Face