Notes of class COMP261 lecture period of 31 January 2002 Background: The project for COMP261 is to build a gadget to help Jason Morris, a blind graduate student conduct archeological field research with other members of a team. Secondarily, the device should be generally useful for him in daily activity. Of all the over 20 wish list features he presented to the class several weeks ago, we selected his top three priorities, and explored technologies to implement them. These were rangefinding, direction sensing, and position sensing. We settled on ultrasonic for rangefinding, magnetic compass for direction, and GPS for position. The lecture period for 31 January 2002 was devoted to a visit by the client Jason Morris to answer more specific questions to help us define the project parameters. Another objective was to try a first prototype of the ranging technology to gain insights, especially if the ultrasound would bother Jason's guide dog Annie. Prototype trial: The prototype was lashed up from a Polaroid developers kit with its large electrostatic transducer and a battery, operated with the default kit parameters. Distance was displayed on the kit's LCD. We followed Jason's initial idea of range readings only on request, implementing it by his verbal request being responded to by Jeff Townsend who read the display back to him. We relocated to the grassy space on the west side of Phillips Hall which has some trees, a nearby lamp post about 6" dia. and is flanked by Phillips on one side and a low stone wall along the driveway on the other, perhaps 20 meters between them. After first determining that Annie did not even seem to notice the sound even at short range and in the direct beam, the initial question seems resolved. Even under these conditions, Annie seemed more interested in other things going on around her. Then Jason tried using the ranger to get a picture of his environment by selectively pointing the device in various directions and requesting readings. It soon become apparent that two effects conspired to make this mode nearly useless. First, the sparseness of sampling caused him to miss sensing presence of some trees altogether when his samples bracketed but did not hit the tree. The other effect was a long term inability to repeat a measurement in the same direction. It looks like incremental angular proprioception works to some extent, but long term absolute angular proprioception is very unstable at best. Subsequent discussion: Accordingly, a different usage mode was proposed which Jason seemed to think was promising. That is, a continuous ranging mode with range being reported by a modulated tone. For example, tone frequency being proportional to reciprocal distance. In addition, an on-demand mode which would report a precise distance, for example by synthesized speech. The first mode would be useful in building a mental image, while the second would allow quantitative measures useful for archeological work. Jason envisions being able to stand at a (presumably logged) way point and measure the orientation and distance to various points of interest, so their geometric relationship could be recorded and their absolute locations estimated. The question of compass accuracy naturally came up. For just a rough direction guide in getting around, 20-30 degree accuracy might suffice. For quantitative work, better than 5 degrees would be highly desirable. In fact, the compass should be able to resolve angles to at least the rangefinder's angular precision which apparently varies with distance and appeared to be a few degrees at moderate distances. Of course selection of transducers and frequency can be used to tune the beam width, so the compass should be as good as the best we may choose for the rangefinder. This also suggests the compass should be co-mounted with the rangefinder rather than separately attached to Jason's body coordinates (belt mounted?) as he had originally conceived. We explored the question of haptic vs auditory feedback to some depth. Jason feels quite definitely that a braille interface is not a good idea, not only because of power, size and weight, but also the limited information bandwidth, the competition for use of his hands for other things at the same time, and frankly, his tactile endurance for extended periods. The tactile compass rose interface was described to Jason and he found the idea intellectually appealing, but again the competition for hand bandwidth seemed more important. When asked specifically to compare the tactile vs. an auditory compass report, he replied, "Let me ask you: would you prefer to have one of your hands tied up?" The idea of a head mounted unit rather than hand held was suggested to Jason, and after thinking it through, he really warmed up to the idea. Advantages include free hands, a feeling that proprioceptive "straight ahead" would be more accurate with head than hand, and a monaural earphone (which he has always wanted) would not need an extra wire. The question of battery life and the trade-off with cabling was discussed. Jason feels having a single cable from a belt pack (NOT a backpack, he has one of those already, thank you!) to a head mount would be far preferable to carrying around enough battery power on his head. Of course an implication of this is the belt pack could also carry all circuitry not absolutely necessary on the head, and probably is a better place for the input interface which at this point looks like push buttons. We learned far more than I had any reason to predict during this session, which lasted until well after the formal class period. This on top of the first class session in which he described his wish list and priorities. My take home lessons from this session are significant: 1. An early prototype in the hands of the client provides design insights to both the designer and the client. Consequently, the specification can be modified to more accurately reflect the client's problem. 2. Involving the client interactively as the concepts are developed provides as much as if not more value in early design decisions as an initial problem statement from the client. 3. I believe we need to repeat this experience soonest with a compass lash up prototype; then with a GPS prototype as soon as we can decide which one to use. Recorded by Leandra Vicci, 1 February 2002