a1174@mindlink.UUCP (Peter MacDougall) (11/26/90)
Perhaps I am entering this discussion a bit late, we just started to get this group and I was away for a week. One of the big questions seems to be how to translate the position of a human body in real space into a construct in virtual space and then to relay the orientation in virtual space back to the person in real space. Right? The human body contains several mechanisms to identify the position of various body parts. The information from these mechanisms is then organized by the neurons of the brain (roughly equivalent to parallel processors for the sake of this argument) and is translated into a construct in the consciousness of the person (the consciousness being equivalent to a virtual serial processor maintained by the parallel processors) Now these mechanisms are: sight: horizon location, bifocal cues, focussing distance cues, and other visual cues such as overlap, atmsopheric changes with distance etc... vestibular: semi-circular canals for determination of rotational acceleration, and the saccule and utricle for determination of orientation of the head with respect to gravity (in the baby the head is strictly oriented with respect to the body simply by muscle proprioception as described below) muscle: there are sensory organs imbedded in the muscles which relay back inforation on i) rate of stretch of the muscle ii) absolute length of the muscle (stretched or not stretche) tendon: similar to the muscle, there are sensory organs in the tendon which determine the absolute tension on the tendon and muscle. skin sensory organs: there are a variety of sensory endings located in the skin to detect temperature, light touch, vibration, and tissue damage Since this system seems to work so well, it probably is a good place to start in terms of modelling mechanical means of identifying position and rate of change of movement of the body in real space. I suppose this is obvious and the question is HOW to do it. Physiology text-books are good for describing the mechanism of how the body does it: ie: how the body converts the the rate of change of length of a muscle into varying frequencies of impulses in a neuron. In a most general sense, one could start a virtual reality simulation with the person in a pre-specified stance, such as at attention, and using this original position, each identified change in position could be added on to the previous position to identify the current position. The change in position could possibly be identified by the change in lenght of a series of thin wires stretched in pairs on opposite sides of each major joint. Take for example the elbow, a simple hinge joint: one wire stretches across the inside of the joint, much like the biceps does, another stretches across the back of the joint, like the triceps does. The length of the "biceps" wire in relation to the "triceps wire will identify the angle of the joint. For ball and socket joints, there would have to be two pairs of wires, one pair for each axis. For modelling head position, one could do as the baby does and simply monitor the head position in relation to the torso via the stretch of the neck muscles. The spine could simply be modelled as a ball and socket joint with two pairs of wires, one pair in the x-axis one in the y-axis. One wire would stretch between the occiput (back of head) and the spine, its mate from the chin to the sternum (centre of chest). The other pair would stretch from the mastoid (just behind the ear) to the acromion (shoulder) on each side. Now for ROTATION, all 4 wires would be stretched, but the normal relationship between each pair of wires would be different than for simple movement in the x and y planes such as with touching your chin to your chest or touching your ear to your shoulder. In planar movements, the relationship between the each wire in the pair is inverse; one stretches, the other relaxes. In rotation, this simple inverse relationship would be constantly changing: the amount of stretch of one wire would not remain in a constant inverse proportion to its mate. So it would be possible to identify rotation. A simple mercury switch located on the forehead and orientated left to right could be used to identify if the rotation of the front of the face was to the left or the right. If the wires were marked with light and dark alternating patches, an optical device could even measure the rate of stretch of the wire much like a desktop mouse's speed is measured. This is very long and I am not sure I have not lost myself or missed something obvious. However it is another idea for getting around the mess of identifying head orientation. Pity it isn't elegant. Peter MacDougall. contact me at Peter_MacDougall@Mindlink.uucp