CSE 492 R - Winter 2020
Week 5: Perception Robotic Vision
Maru Cabrera
February 6th, 2020
Agenda for Today
Demo presentation: WITH discussion
Mention: description of the demo, challenges, future aims
Introducing the topic of Perception or Robotic Vision
Reading Assignment 2 is out! Due next week (Feb 13th).
Demo 1 Presentations
Indoor Navigation Using the Fetch Robot in Simulation
Summarize or Describe your group’s Demo
What were some of the challenges and how were they addressed?
How could this demo be potentially expanded in the future?
In robotics, perception is understood as a system that endows the robot with the ability to perceive, comprehend, and reason about the surrounding environment. This allows the robot to make decisions and operate in the real world.
Perception or Robotic Vision
Environment
Perception
Action
Planning
Perception or Robotic Vision: Examples
Robotic perception informs decision making for actions taken in the real world.
Interacting with objects and people in the environment.
Perception or Robotic Vision: Examples
Detecting what is foreground and background in an image
Perception or Robotic Vision: Examples
Importance of Geometry in Image Projection for Augmented Reality
Perception or Robotic Vision: Resources
What is robotic vision? Why is it hard?
Morovec’s paradox:
What humans find hard, computers find easy and vice versa
Lack of generalization
Concept of Visual Intelligence
TIME FOR HANDS-ON WORK!
Login information for workstations:
username: robotics
password: AX2020wi!
Reading Assignment 2
Reading Assignment 2 is out today: Individual assignment. It will be due next week (Feb 13th)
“Intelligent Robotic Perception Systems”
Read up until you reach Section 4: Case studies
Submission will include a written report (max 500 words) where you describe the Homeskies Project case study and how robotic perception will be part of that project.
Looking ahead
Next Week:
Continue with Perception
Object Recognition - YOLO (“You Only Look Once”)
Familiarizing with Intel’s RealSense
Merging these perception components with ROS