Multimodal User Supervised Interface and Intelligent Control

IJCAI [International Joint Conference on Artificial Intelligence] Impressions

1. Workshop

As you may know, I sent in a paper [1] on the Multi-modal project at the workshop "Developing AI Applications for people with Disabilities", held in conjunction with the main conference. [2]. I returned to work officially today.

There was a very interesting mix of papers on mobile robots, intelligent wheelchairs, man-machine interaction, and natural language processing. The papers from ASEL [3] were very well received and I had some very fruitful discussions with some of the organizers and participants.The AI community is going through an intensive soul searching right now, in trying to define it's goals. In fact, one of the invited papers during the paper was entitled, "Turing Test Considered Harmful" by Hayes et al. and focused on how the community has appeared to be lost in a goal analogous to the ancient alchemists. Their search [4] ultimately matured into modern sciences like chemistry and physics. And AI too must mature and redefine what it's goals are and towards practical meaningful applications of AI which could also act as a validation of AI itself as a science.

This idea also permeated the workshop, where rehabilitation research was considered to be an ideal domain for the "true test" of AI. The term that was much bandied about during the workshop was "Cognitive Prosthesis". AI acting as an aid to supplement and complement areas where the user may be deficient and vice versa. In fact, John Aronis's [Organizer, Pittsburgh] summary of how the human and AI system should be integrated appeared to reflect one of the underpinnings of our project.

1.2 Highlights

1.2.1

A publisher is interested in picking up the workshop proceedings and publishing it as a book. John Aronis spoke to me about it and he was interested in getting more papers from ASEL for the proposed book [5]. The workshop consensus was that it was imperative to bring the AI Academic community and the Rehabilitation community together for mutual benefit. In this context, it was decided to explore having a regular panel session for the next IJCAI [Sweden, 1997], so as to bring more AI people into looking into the rehabilitation domain.

1.2.2

The workshop consensus was that it was imperative to bring the AI Academic community and the Rehabilitation community together for mutual benefit. In this context, it was decided to explore having a regular panel session for the next IJCAI [Sweden, 1997], so as to bring more AI people into looking into the rehabilitation domain.

1.2.3

We appear to be right up there with the best of them in terms of the work we do here at ASEL.

1.2.4

In this context, I was thinking whether we should take a more active part here. We could organize a similar symposium, as part of the 1996 AAAI fall symposium series. The call for proposals have been published and the deadline is October 15, 1995.

http://www.aaai.org/Symposia/Spring/1996/sssproposals-96.html

The multi-modal project has a paper which will be presented during this year's fall symposium, and the call for papers for the 1996 spring symposium. There is a session on "Planning with Incomplete Information for Robot Problems",

[http://www.aaai.org/Symposia/Spring/1996/sssparticipation96.html#RTFToC24

which may be an ideal forum to highlight the multi-modal project's planner component. [6]

2. Robot Events

A robot competition is held at every IJCAI. This were there was a new exhibition on intelligent wheelchairs.

2.1 Competition

The competition involved testing two separate activities. One involved maneuvering around a office space, which involved recognizing doors [both open and closed], hallways, and mainly involved moving from a starting room to a target room. The other competition involved picking up trash [cans and paper cups], and then putting the cans into a marked recycling bin and the paper cups into a trash bin. My impression was that, we could have spruced up our Mechatronics course robots and would have given everyone of them a run for their money.

2.2 Intelligent Wheelchairs

There were three wheelchairs in exhibit. The Penn walking wheelchair with which we are very familiar with. I was quite pleased when they acknowledged both AI Dupont and ASEL during their presentation. It seems that there funding runs out August 31st and the whole project is going to be put on hold.

The other two wheelchairs [Navchair from Michigan, and Wheelesley from Wellesley] used sonar sensors to navigate around obstacles in a room. Navchair was exhibited at this year's RESNA.

3. General Impressions and Suggestions

The work we do here at ASEL is at par with the kind of work I saw during the conference. In many cases, we may be ahead of the "competition" in how we are applying AI in trying to solve *real* problems.

We should participate more in such academic programs. This way, we could get more people involved in our challenging and *useful* field. Publicity can always help, not harm us.

I would like to see us take a leading role in such academic events, and as starters we could investigate organizing a symposium next fall in conjunction with AAAI.

Footnotes

1. "Multimodal User Supervised Intelligent In terface and Control for Assistive Robots.
2. The Workshop proceedings are on my desk.
3. Peter Vanderheyden, Chris Pennington and Cathy McCoy.
4. Transmuting lead into gold; The universal solvent.
5. American Elsevier.

Last Updated: March 5, by Zunaid Kazi <kazi@asel.udel.edu>