Wednesday, July 1, 2009

Mobility perspectives for post-Kalamazoo

For technical reasons please continue by reading the follow up comments to this post.
Thanks Gene

5 comments:

GB said...

During the meeting of 2009 June 29 at Touch Graphics we had the opportunity to review the information gleaned in Kalamazoo. New approaches to monitoring the cane movement seem promising but crucial aspects of cane manipulation behaviors still cannot be captured by the remote.

My concerns relate to the accuracy and completeness of the data that creates the feedback, and insuring that the behaviors we shape with the feedback conform to traditional and/or best practices in the O&M field. Steve has a great list of aspects of mobility behaviors we would want to be able to improve which we can use as a check-off list of desired capacities.

We discussed a proposed design with a row of horizontal light sources at the end of a course that are visible to the wii camera. Assuming that the subject/student begins the course (30 to 60 feet in length) squared-off and aligned to walk a straight-line path, this configuration should be adept at detecting veering. It should also easily detect the cane’s lateral movement and capture (with changes in acceleration and with light) the measure of the cane’s arc.

The new design appears to easily monitor for feedback veering and arc width. But there are two problems or limitations with this approach which must be dealt with.

1) The calculations currently assume two things that we cannot assume: the hand is centered and the arc movement is created by wrist, not arm movement. We should be able to accurately capture cane movements regardless of how the cane is moved.

2) What it does not do, and what we would prefer it to do, is “see” the cane in relationship to the body. Assuring an arc width based on the widest part of the body is only appropriate if the cane is centered. When the wrist or arm is not centered (like when the user holds the cane “holster” style or with a dominant-hand-bias) then a perfect arc width will be create over-coverage on the dominant side and under-coverage on the non-dominant side. Ideally we should create feedback that insures, no matter where the hand and wrist is placed or moved, adequate coverage on the left and right.

Items of Steve’s list of behaviors to capture include in-step/in-rhythm cane movements and other aspects which I think would be best left for future development as we enhance the wii cane with each subsequent iteration of the device

Bonnie DB said...

Steve -Thanks for your through report. I think I have a BASIC understanding of the concerns. Is my brief summary correct?
1. Purpose - to compare user's movements as measured by WiiCane with Optotrak motion capture apparatus.
2. The 12 inch gap between lights on the strip were too far apart so rearranged them to be 6 inches apart staggered pattern. Not a big difference but better.
2. The cane shaft blocked the WiiCane's camera from finding the lights so the Wii was mounted under the cane, which helped somewhat. Still hard to keep track of lights when cane swung out too wide to either side.
3. Next, Wii was angled to point almost straight down to floor. Did not lose sight of lights when cane arc at its widest point.
4. Decision to end test because need to rethink and reprogram geometry between camera and lights.

3 POSSIBLE SOLUTIONS
1. Use light bar at end instead of strips.
Advantages - installation easier, traditional Wii gaming system does this.
Challenges - Wii camera is very far away from light source - 30 ft. Require mounting Wii on top of cane in holder at level angle. Throws balance of cane off. Adds cost. Complex.

2. Add fisheye lens to Wii. Must maintain view of 2 lights on strip at all times.
Advantages - more lights in view when cane swings wide.
Challenges - Find lens & hope it is inexpensive.

3. Make adjustable mount that permits Wii to hang down & point downward.
Advantages - May be able to optimize field of view to permit acceptable measurement of cane movements.
Challenges - cost, time needed to engineer mounting device, reduces overall length of ellipse.

STEVE'S RECOMMENDATION
1. Create adjustable fixture for mounting Wii tilting downward to point to lights on floor strip. I am not familiar with "dampened gimbals" but it seems to mean that somehow the device would always hang down. Does this add weight to device?
2. Look at approaches for refining design of light strip.

QUESTIONS -
1. Do we need to know where cane is at mid-swing in order to measure height tip is from floor?
2. How does position of Wii mounted on cane - above, below, facing forward or downward - effect balance of cane & fatigue. Please remind me how much Wii weighs. Would Wii weight have effect of making it less likely cane would be lifted high off floor in center of arc?
3. Why do you think the fisheye isn't a viable solution?

FINAL COMMENTS - I think it is OK that if the cane is used "wildly" it may be impossible to maintain awareness of it. In my mind, this device would best be used after initial O&M instruction. You wouldn't want a student who doesn't have at least a general understanding of arc width and height to be using the Wii for feedback/reinforcement. Agree with Steve's idea of series of feedback given if person's cane leaves the measurable parameters.
Bonnie

Bonnie DB said...

Gene - As you say, we need to know where the cane arc is in relationship to the person's body no matter whether the cane hand is centered or how the arm/wrist is moved. I agree with your statement "We should be able to accurately capture cane movements regardless of how the cane is moved."

The bottom line is whether the cane user's body entire body width is covered and how high the arc is off the walking surface.

I don't understand how a row of horizontal light sources at the end of the course can measure veering at the beginning of the course, due to the 30-60 foot distance. Steve's report seems to state that using light strips the entire length of the path is best. I'm interested in knowing why the light at the end was proposed.

Bonnie DB said...

Oops - I meant thorough report - not through report Steve.

sl said...

Bonnie, your summary of where we are is succinct and excellent. You pretty much captured it. I don't have many answers for you, except to say that I believe that our mounting idea will give us a lot of flexibility to try different light configurations very rapidly so that we can fairly quickly arrive at the best approach. The wii device is quite light: with batteries it weighs about 5 ounces. the fixture adds a couple of ounces, but we are cutting out a section of the cane, so we lose one or two ounces there. I am guessing that the cane will feel noticeably heavy, but not so much that it affects performance. But, rather than speculate, people who are at the meeting tomorrow can try it for themselves.

Regarding your question about having to know where the mid point of the cane arc is to determine the height of the cane tip above the floor: again, I don't know, but I think it will be relatively easy to determine the the height of the cane tip above the floor at any moment. So, if we know the mid point of the arc, we can get the height at that moment.