Overview
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore cutting-edge advancements in robot perception systems and algorithms in this Stanford seminar featuring MIT's Luca Carlone. Delve into the challenges and breakthroughs in spatial perception for autonomous systems, including self-driving cars and unmanned aerial vehicles. Learn about high-level 3D scene understanding, metric-semantic mapping, and novel hierarchical representations like 3D Dynamic Scene Graphs. Discover Hydra, a groundbreaking real-time spatial perception system that builds 3D scene graphs without human supervision. Examine the connection between robustness in robot perception and global optimization, and understand certifiable perception algorithms that offer performance guarantees even in extreme noise and outlier conditions. Gain insights into vehicle pose and shape estimation for self-driving scenarios, and explore the theoretical implications of these innovative approaches to robot perception.
Syllabus
Introduction
Spatial Perception Systems
KIMAERA
Chimera
Dynamic Sync Graphs
Chimera VSG
Hydra
Optimization
Demonstration
Audience Question
Recap
Perception Problems
Robust Estimation
Certifiable Perception Algorithms
General Philosophy
Proof
Lasser moment relaxation
Perception toolbox
Open problems
Wrap up
Taught by
Stanford Online