Tutorial at ICRA 2018, Brisbane

May 21, 2018

Tutorial organizers:

Anelia Angelova (Google Brain / Google Research) anelia/at/google.com

Jana Kosecka (George Mason University / Google Brain) kosecka/at/google.com

Presentation slides

Part 1 - Introduction

Part 2 - Deep Networks for Robotic Vision

Part 3 - Applications: Segmentation, Pose Estimation and Matching

Part 4 - Applications: Grasping and Learning from Videos

Tutorial schedule:

9:00 - 9:05 Welcome and overview

9:05 - 9:30 Intro (Jana)

- Motivating examples

- Perceptions Problems

- From handcrafted features to feature learning

- Earlier and current pipelines

9:30-10:30 Perception for Vision and robotics (Anelia)

- Classification CNN architectures, Inception, Resnet, Densenet

- Classification applications in Robotics

- Detection architectures: Faster-RCNN

- Detection applications in Robotics

10:30-11:00 Coffee break

11:00 -11:45 Applications Part 1 (Jana)

Pose estimation and matching

Scene Segmentation and Dense prediction

11:45-12:30 Applications Part 2 (Anelia)

Grasping

- Grasp detection

- End-to-end and self-supervised grasping

Learning from videos

- Temporal Learning

- Self-supervised learning of depth and ego-motion