dietz@SLB-DOLL.CSNET (Paul Dietz) (02/02/86)
The Challenger explosion has got me thinking about more about teleoperated mechanical manipulators for space tasks. These are robot arms or similar devices controlled from Earth by humans. Such devices already exist on Earth as manipulators on submarines and for handling radioactive substances. The problem with using them in space is the nonnegligible feedback delay imposed by the speed of light: about .25 to .3 seconds for geosynchronous orbit and about ten times that for the moon. There are lots of interesting questions raised by these machines. How inconvenient is the feedback delay? What kind of force sensing is needed to prevent the manipulator from crushing things? What kind of sensors should the manipulator have? Cameras, surely, but how many and what kinds? How much image processing should be done, and how a should the resulting information be presented to the operator? Where should the processing be done: in space or on the ground? The remote manipulators would have many uses, and would probably be built with interchangable effectors for various missions. Suggested missions include: satellite refueling and/or part replacement in geosynchronous orbit, lunar mining and manufacturing, and orbital manufacturing of habitats or solar power collectors. Gerard K. O'Neill estimates that, using remotely controlled manipulators, a "seed" manufacturing facility capable of making 1800 tonnes of material per year could be put in place with 107 tonnes on the moon and 89 tonnes in high orbit. The facility would be capable of reproducing itself (except for some "vitamins", like integrated circuits and some volatiles) in 90 days. Remotely controlled manipulators probably also make sense even if humans are in orbit. For example, it probably would be more efficient (and safer) for the human to be close (< .05 light seconds) to the manipulator (and inside a habitat) than for the human to be in a space suit. A study of the effect of feedback delays on manipulator performance would seem to be an excellent research area for a robotics lab. This research would probably be the single most important near term contribution a computer scientist/roboticist could make to the space program.
rjnoe@riccb.UUCP (Roger J. Noe) (02/05/86)
In article <8602020218.AA04732@s1-b.arpa> Paul Dietz writes: > The problem with using [remote manipulators] in space is the > nonnegligible feedback delay imposed by the speed of light . . . Well, that's ONE of the myriad problems. What about maintenance and repair of the machines? Or ordinary servicing, including refueling (if any)? Another minor problem [sarcasm] is the design and realization of these machines. Certainly we should be able to replace humans with teleoperators in every dangerous "mundane" occupation (e.g. coal miner, test pilot, saturation deep-sea diver, fire fighter, police officer, cat bather, and inner-city school teacher to name just a few) long before we'll be able to create teleoperators that function even in low Earth orbit. I guess it'll be a while before we even see such things happening on the surface of the Earth, much less in space. > . . . a "seed" manufacturing facility . . . could be put in place . . . > The facility would be capable of reproducing itself . . . What, you just drop it on the lunar surface and it installs itself? That would be multiplying the complexity of the thing considerably. And of course you build in a Junior Alchemy set so that it can transform lunar soil into titanium. Oh, not impossible maybe, but certainly a couple centuries into the future. And many quadrillions of dollars. > . . . it probably would be more efficient (and safer) for the human > to be close (< .05 light seconds) to the manipulator (and inside a > habitat) than for the human to be in a space suit. No argument there, but it does pose some technological challenges in the areas of dexterity, adaptability, etc. > This research would probably be the single most important near term > contribution a computer scientist/roboticist could make to the space > program. You have a rather elongated definition of "near term." How about a computer scientist becoming an astronaut to understand better what needs to be done with data management systems to better assist people in space? That's my goal. -- Roger Noe ihnp4!riccb!rjnoe
henry@utzoo.UUCP (Henry Spencer) (02/06/86)
> > . . . a "seed" manufacturing facility . . . could be put in place . . . > > The facility would be capable of reproducing itself . . . > > What, you just drop it on the lunar surface and it installs itself? > That would be multiplying the complexity of the thing considerably. > And of course you build in a Junior Alchemy set so that it can transform > lunar soil into titanium. Oh, not impossible maybe, but certainly > a couple centuries into the future. And many quadrillions of dollars. Actually, it's not very hard at all, or terribly expensive, *if* you send along a handful of humans to do the hard parts. This problem has been studied. It is not terribly difficult to do 90% or more of the work by remote control or automation. The remaining 10% or so is vastly harder to handle that way, and the most economical approach is to use humans to fill in the gap. If the human end of the operation is a colony rather than a base (key difference: colonists don't expect to come back), it is not particularly expensive. -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry