by Dan Lester
|If humans could get close to the planetary surface, perhaps just in orbit above it, truly cognitive teleoperation could be achieved at multiple sites with high-quality virtual presence.|
This vision seriously colors the public view of space robotics in ways that are less than constructive. While Spirit and Opportunity move, see, and can even touch, they conform only loosely to that public view of robots. Cassini can move and see also, but the popularization of the word “robot” fits it even less well. It doesn’t have legs, arms, or even wheels! Certainly Robonaut 2 on the International Space Station is widely appreciated as a robot, because although it hasn’t really done much yet, it has a head and fingers, and it can even sign a greeting and shake a hand!
As noted in my earlier essay (see “Human spaceflight, and the reason for (almost) being there”, The Space Review, July 5, 2011), using machines to bring our cognition to new places in space is an effort seriously handicapped by distance, the speed of light, and resulting communication delays. When we think of such machines operated from Earth, they involve some level of autonomy. We upload an instruction set and a task plan, tell the machine to please be careful, let it do what we tell it to do, and wait patiently until it reports back. It’s not unlike a parent-child relationship. It is in this way that human exploration looks comparatively attractive in that, by physically going somewhere, humans can really experience a venue in a fully cognitive and, shall we say, adult manner, on time scales consistent with the human reaction time.
For space applications at a distance (Mars, for example), control from the Earth involves many minutes of delay that make teleoperation highly non-cognitive. As a teleoperator, it’s hard to feel like you’re really there, though Mars Exploration Rover operators have claimed to be able to do so at some level. But if humans could get close to the planetary surface, perhaps just in orbit above it, truly cognitive teleoperation could be achieved at multiple sites with high-quality virtual presence. This strategy, of space exploration by telepresence, was envisioned independently by Fred Singer and Marvin Minsky in the late 1970s, a time at which our technology simply wasn’t up to the task. It is now.
In this way, much as for the teleoperation we do now with machines on Earth—in mines, on the deep sea floor, and even for surgery—the machines hardly require any autonomy. These “robots”, as we call them, can be just tools that extend the senses and dexterity of a human to a particular site, which may be distant. In this way, automobiles are telerobots that we all use, as we sit in the drivers seat. With a steering wheel, levers, and pedals, we remotely control the actions of the wheels underneath. Not exactly control at a distance, but not conveniently close. We watch the gauges, as well as look out the window, to sense the performance of the vehicle. Bulldozers are telerobots that move earth in response to commands given from the cab on top. These are tools that extend (and can even magnify) our dexterity and strength. Telephones are tools that extend our auditory presence. Are they really “robots”? Certainly not in Issac Asimov’s model. I hardly worry about the morality or faithfulness of my car. But these machines are physical extensions of the person using them, who may well say “I, Robot”, with some accuracy. The “I” is that controller person, whose cognition and dexterity is just being transmitted somewhere else. It isn’t the machine, which is just a tool in which that cognition and dexterity is manifested.
|The word “robot” can be distressingly imprecise.|
In this short essay, it is worth touching on a recent paper in the journal Astronomy and Geophysics titled “Dispelling the myth of robotic efficiency” by Ian Crawford at Birkbeck, University of London. Crawford has chosen to give new breath to the exasperatingly archaic humans-versus-robots argument about space exploration. He concludes, somewhat differently than the view of physicists Bob Park and Steve Weinberg, that humans are vastly better than robots for doing space science. Crawford does add somewhat to the discussion by attempting a real value assessment, as in science-per-dollar, though his assessment is debatable in many respects. His definition of “robots” is clearly restrictive, as in robots without astronauts, and robots controlled from Earth. In the Singer and Minsky picture of on-orbit telerobotics, such a view is unreasonably narrow. A different question about robots and space exploration is perhaps more germane. Given that human cognition is highly advantageous to space science, what is the best way to bring that human cognition to an exploration site? It may be that machines can be a big help.
In these ways, the word “robot” can be distressingly imprecise. To a layperson, the word connotes some level of intelligence that, I suppose in a very loose and generous way, my automobile might just have a little bit of. But the machines that we will teleoperate on distant planets, ideally from a control station suitably close by, will be remote control tools: cameras to help us see, haptic sensors to help us feel, and the equivalent of prosthetics to help us manipulate, rather than machines with any sense of “self”. Now, some level of autonomy might still be valuable in improving their operational efficiency, and certainly our powers could be expanded, but the zero-order architecture is one in which these units are simply extensions of… us.
Perhaps we shouldn’t call these things robots at all, with all the Asimovian baggage that word carries. Maybe “cognibot,” which transmits our cognition to another place, or “humabot,” which brings our humanity to a new venue. Perhaps “telepresences,” as suggested by futurist/ideonomist Pat Gunkel. Maybe just “electromechanical puppet” or, even simpler, “marionette.” But when we talk about achieving tasks in space with low latency control, with modern sensors and high-dexterity haptically-enabled manipulators, let’s not fall back on the tired excuse, “Oh, but robots can’t do that.” Because these aren’t robots. They’re us. One step removed.