MINSKY@MIT-OZ@sri-unix (09/08/82)
From: Marvin Minsky <MINSKY at MIT-OZ> The effect of time-delay on remote manipulation will depend on the time-scale of what you are manipulating. The human sensory-motor reaction loop is of the order of 1/5 of a second. As you know, things fall 16 feet in the first second, and there's a square-root scaling for that. So you can catch something that falls out of your hand in half a foot or so. The lunar gravity is about 1/7th ours. The lunar round-trip time delay is about 12/5 seconds, and if we add our own delay we get 13/5. So if you consider gravity scaling, then lunar teleoperation should be sort of like you are twice as slow as here! That is, you ought to be able to catch something before it falls two feet. Now, we will be building lunar structures that are large, because of the low-G. Perhaps they'll be 4 times as high, and yet use weaker materials! Then you can catch them just the same as here - when you scale everything! How's that for an optimistic way to look at it? Seriously (but I was, actually), most space-structure work will use very delicate materials. That, after all, is the great thing about building stuff out there. Then, the manipulation speeds will be limited, not by reaction time but by the gentleness needed for handling. You simply cannot accelerate those aluminum-foil girders very much or they will bend and crinkle. So space teleoperation will not suffer from delay very much, because you will have to plan your motions many seconds ahead, or even minutes, to avoid collisions. If you jerk to avoid an unexpected collision you might do more harm than good abyway. Finally, of course robotic AI systems can work fast locally. If you want to take your analog wrist-watch apart it will take a long time, by remote control, but presumably such tiny work will be the exception. Epilogue: it infuriates me that there seems to be still no substantial research on telepresence. There are a few little projects here and there, but none of much significance. Still, no one seems to be fabricating a decently humanoid remote hand, either. Foo on the U.S. if some other culture beats us on this, another of the obviously enormous industries of the not far-away future that we will miss out on. -------
REM@MIT-MC@sri-unix (09/09/82)
From: Robert Elton Maas <REM at MIT-MC> Note that I was not proposing the use of 3-sec-delay remote-control for fabricating aluminum girders, merely for stuffing ore into a hopper for tossing out to space (with or without preliminary processing on the moon before the toss). Thus the problem of fragile materials and "dropping things" won't exist. Once ore gets off the moon it's trivial to move it to LEO where there are plenty of voluteers to handle it locally. The reason for suggesting telepresence on the moon is it's a real pain to maintain human crews there, whereas LEO and even lunar orbit are easy by comparison. The reason for suggesting telepresence between Earth and Moon instead of between lunar-orbit and Moon or LEO and Moon, is that telepresence may be very clumbsy, taking lots of time even to do loading of ore, so using an astronaut's time would be cost-ineffective, whereas using the time of an Earther would be reasonable, even with the 3-sec problem. A lot of out-of-work Earthers could be employed as remote manipulator miners. (Hey Reagan, did you hear that? A solution to unemployment!)