March 2007


The Semantic Robot Vision Challenge is a new research
competition that is designed to push the state of the art in image
understanding and automatic acquisition of knowledge from large
unstructured databases of images (such as those generally found on the
web).

In this competition, teams will be required to demonstrate a robot
that has the ability to:

  1. Autonomously connect to the Internet and build an object
    classification database sufficient to identify a number of objects
    found on a textual list.
  2. Use this classification database to autonomously search an indoor
    environment for the objects in its list.

Integrating a mobile robot with the vision research adds another
interesting layer of complexity that would not ordinarily be available
in a purely computer vision competition. Semantic understanding of
the objects could also be expanded to scene understanding as well.
Thus, scene context can be used to guide the search for objects in
areas which make the most sense for them to be found (e.g. a stapler
is usually found on a desk rather than on the floor or on a wall).

The Semantic Robotic Vision Challenge

Technorati Tags: , ,

(reprinted from http://jerryandsarasteele.typepad.com/deez_steeles/2007/03/hri_2007_confer.html)
I attended the 2nd Annual International Conference on Human-Robot Interaction (HRI) this weekend in Arlington, VA. I presented a poster titled Directed Stigmergy Control of Multi Robot Systems. This is the first academic conference I have attended in about 7 years, and I think it was a nice conference to get back into the mix. I met a lot of (sometimes intimidatingly) smart people, both faculty and fellow students.

  • I was really struck by the diverse interests of the conference patrons. I think this diversity reflects the wide range of disciplines that makes up human-robotic interface research. There were people with psychology and social psychology backgrounds, people with music backgrounds, and of course computer science and engineering. I enjoyed the diversity because it challenged me to think of robotics and the impact of robotics on society much more broadly than I had. In our research we had focused on scientific robots, particularly mobile robots used to explore Mars. This conference brought to my attention many more kinds of robots: robots to assist the elderly and blind, robotic shopping carts, robots that interact with children and give guided tours of museums, that play drums interactively with other drummers, robots that act as proxies for individuals physically located elsewhere.
  • Much of robotics research is taking place in Japan, and there were a lot of Japanese attendees. Some of their presentations blew me away. Japan seems to be making a national investment in using robotics to change society, both as an assistive technology for the disabled and elderly. One of the keynote speakers from Japan, Hiroshi Ishiguro, had a really interesting perspective on robotics. He believes that the study of robotics is the search for deeper understanding of human behavior. His goal is to understand how humans behave and interact with each other to such a detailed level that he can create robots that behave like humans, and allow humans to interact naturally with the robots he creates.
  • One of the recurring conversations at the conference was trying to understand exactly what is Human-Robot Interaction as a scientific discipline? Certainly we draw from engineering, computer science and computer vision. As we seek to make interactions between robots and humans more natural, we draw from more social sciences like psychology, social psychology, even neurobiology. But it appears that HRI has yet to really crystalize or formalize as a discipline in and of itself, in that there are no central theories or agreed upon principles that we all agree upon. It may be difficult to come to that kind of consensus, since right now we build robots with particular traits to solve particular problems…its hard to generalize techniques for a Mars robot to techniques for a robotic shopping cart. At any rate, it is kind of exciting to be working in such a young field. Lots of opportunity, even if there is some risk that in 5 or 10 years it may not exist.
  • I thought it was even more fun to ask the question, “What exactly is a robot?” One definition is that a robot is a collection of automation that senses the environment and can take action on behalf of the operator. So is the automatic flushing toilet a robotic toilet? I think if you asked 50 roboticists, you’d get 50 different answers.
  • Central issues in HRI: People tend to anthropromorphize (treat as though it were human, with human intelligence and motivations) robots. This is a problem because we don’t yet have the artificial intelligence to make the robot behave completely natural, so people get confused. So a lot of the papers were sort of demos of making robots interact with people in different ways and different situations, and seeing how they react. One paper made people play with a little robot, then asked them to kill it by smashing it with a hammer. People said things like, “poor little guy, ” or “that’s so inhumane.”

One thing that you have to consider as a budding academic is which societies and associations you want to align yourself with — whose papers will you read and who do you want to read your papers. I think I will continue to be involved with the HRI crowd because I feel its diversity is a strength. Robotics can take you into so many fields like space exploration, medicine, and in the home. That gives you a lot of opportunity to collaborate with different people in different fields, and extend your own research in lots of different ways.

Technorati Tags: , , , , ,