Sunday, October 18, 2009

project status and ongoing discussions

First, I need to apologize for not participating in the ongoing discussions that are occurring on email regarding which feedback types are appropriate for young children and also which style of cane use our system will train students for.  As I mentioned, last week  for me was consumed with preparing for and then attending a great conference at the Metropolitan Museum.  Now that's over and my full attention is shifting onto WiiCane for the next two weeks as we get ready to do our first field tests.

I need to reiterate that our group discussions must take place on this blog, so that we end up with a coherent record of our conversations, and so that all project participants can easily keep abreast of our developments and make contributions as needed. I understand that there were some oversights in distributing invitiations to  blog participants, and I also know that it can seem easier or more satisfying to just fire off an email to one or two people, but it is important to the overall project that we use the blog. I expect traffic to pick up significantly now, so it's really a very good idea to subscribe to the email digest that will come to you on every day when there is some blog activity. If you have not signed up for that yet, please do so by filling in your email address in the text box in the upper right corner of the blog and then clicking "subscribe".  If you have not received an invitation to participate in the blog, please let me or Gene know, and we will take care of it.

Regarding the question of how best to provide corrective messages to young children undergoing training with WiiCane: I think that this is a question that will be clarified quickly when we start running kids.  My guess is that directional audio that plays a tone in the ear on the side toward which the correction should be made will be more effective than saying "move right" or "move left".  But, it's easy to create a feedback system that allows for either option to be set, so we should have no difficulty in determining which works best, or, we provide both options and allow the test administrator to decide.

On the question of two-point tap vs. constant contact: my understanding is that most users these days are trained in constant contact, and so we need to ensure that our system reflects this preference. We should not develop a system that is only capable of providing instruction in an obsolete or unpopular method for cane use.  But, the proposal was written specifically to address two-point touch, and so we need to do that first. Once we get that working, we can augment the system so that it can also be used for training constant contact also. From a technical perspective, it is much simpler for us to first develop a tap-based system for motion capture. We know that tap events are very easy for the Wii's accelerometers to recognize. So, our current conception calls for taking a "snapshot" of the cane's position at the moment when a single tap occurs. If we define cane arc as the angular or lateral diplacement that occurs between tap events, our work is greatly simplified. Without taps, we will have to develop a much more complicated approach to arc width calculation, in which we would have to recognize the moment where the direction of cane movement reverses. That's not impossible to do, but it is probably a lot harder. My thinking is that we should master 2-point touch first, then introduce constant contact later.  The next couple of months are going to include a lot of trial and error as we develop our technology and our pedagogical approach. Nothing is fixed in stone right now, and we need to remain very open to making changes and enhancements suggested by our domain experts and our users. but, we should not abandon plans that were approved by the grant review panel without receiving explicit permission to do so. I also think that its sensible to start with something that we know how to do, get that working, and then move on to them more complex stuff.

On Tuesday, Zach and I will get together to connect his Java WiiCane drivers to the test administration program that I am working on. We should have something nice to show on Wednesday, when I am assuming that we will meet at the office to discuss our plans for responding to the IRB's questions about the test protocols we have requested them to certify.

1 comment:

GB said...

Steve, thanks for the update. We needed time to talk this out among ourselves anyway, so welcome back to wii-land.

My concern about the tap/no tap challenge are:
1) we may very well need to elimate many, many potential subjects because they cannot present an organized tap. This is a problem just waiting to happen at the worst possible time -- when we are out there at the sites
2) momentum - why don't we tackle this now. I am sure if you guys put your heads together it can be done - and I would like to be involved.

I can't imagine the funder having any concern for this minor change in the propoed device. After all, didn't we promise a game-like system?

I don't know how to emphasize my concern on this issue beyond saying that I am worride on several levels. In my opinion, getting an approximation of cane-tip-placment now, even if it is imperfect, would be far more beneficial than providing a very accurate tap event.