Robot Bodies Needed Before Robot Minds
It had to happen – Sustainable Robot Design
August 20, 2012Posted by on
If you see my recent article over at .NET magazine on the sustainable web, you’ll see that I feel that sustainability is something that is ignored in the tech world, often in favor of magical thinking about the power of the goddess Techna to rescue us. Considering that sustainability is big in graphic design, industrial design, architecture, even interior design, what about robotics?
For the sake of argument (since this blog is about Robots That Jump) lets consider the restricted definition of sustainability as minimizing energy and resource use. This isn’t the real definition of sustainability, but it will do for the arguments ahead.
First, let’s consider hobby robotics. Here, the “robotic” way of teaching technology and exciting students can be justified in its efficiency. Most robotics works by schoolkids includes small devices run by battery-operated microprocessors. While a computer workstation is typically used to compile programs down onto the hardware, it is probably using less power than watching YouTube or Hulu for the same amount of time, both in bandwidth terms and cpu/power expended terms.
I find it interesting that GeekBeat thinks this is sexy.
Next, onto “battlebot” style rampaging robots. This includes the smallish tele-operated machines trying to destroy each other, as well as the monsters produced by Survival Research Laboratories and their ilk.
And the big boys battling….I like the honesty here. In most tech-stuff, amazing devices like “air cars” are show operating nearly silently. In practice, lots of real-world technology is very, very, very noisy.
Now, for the bad stuff. The sort of articles savaged in the blog typically have titles like:
“Robot you can order to get you a beer ALMOST ready to buy at Wal-Mart!!!!”
And for today’s pathetic example, this lame YouTube video featuring Chinese robots slicing noodles.
Phew. One comment: I’m absolutely sure they aren’t taking over the food industry. Not now, or anytime soon. Quasi-religious hype grafted into endless techno-utopian fantasy.
What about sustainability? If you think of these as research robots, not so bad. To develop better robots, one needs to explore how to make their bodies agile, since artificial intelligence is not likely to be present anytime soon. So what if we have to put our press release lying to the public, or send our stories to ignorant newspeople who will gobble up anything we say? It is in the services of the higher cause of robotics.
But in terms of the ideas spit forth by the video above, we see a crass appeal to the unsustainable. An average adult male burns about 120 watts of power living their daily life, with their brain taking about 25 watts of that power. Olympic athletes, and the guy who flew a human-powered plane across the English Channel, can train themselves up in the range of 1 horsepower (760 watts). These robots can’t even come close to human efficiency, even if we accept that they don’t need real brains yet.
This is really not much power. A successor the Solar Challenger, had 16,000 photocells and could pull in about 2,000 watts in direct sunlight – enough to run two hair blowdryers. The man above couldn’t even run one. It did this in 1981. We haven’t advanced much from that.
Interestingly, humanoid robots like the Asimo and HRP series are using about 150-200 watts during normal operation. But the Asimo can only pick up a kilogram or two of stuff, less than a human child.
Unlike humans, they are horribly inefficient, and can’t win world records, much less even walk smoothly. So, one of the reasons we don’t have Robots That Jump is that they are inefficient. Using this level of robot in the world would make it less sustainable. We don’t have to worry about movie robots anytime soon, which seem to be able to rip office buildings in half without breaking a sweat of grease – they don’t exist, not even remotely.
Well, wait, I saw that robot-donkey from Boston Dynamics!
The truth is, Boston Dynamic can’t really power their robots. The Big Dog has a gas engine roaring away like the mother of all Saturday-morning lawnmowers to keep moving. As for the “Cheetah” above, it uses:
The current version of the Cheetah robot runs on a laboratory treadmill where it is powered by an off-board hydraulic pump, and uses a boom-like device to keep it running in the center of the treadmill. Testing of a free-running prototype is planned for later this year.
Real Cheetahs are quiet when they run…
…unlike the coming metal monsters. Get your earplugs on! A lot of noise, by the way, means you’re wasting energy compressing air instead of moving a robot body.
How does that inefficiency add up? For the present, a human will be able to do the same job as a robot for far less energy. In fact, if it is a job requiring a humanoid robot (like cutting noodles), they will be the equivalent of about 10 humanoid robots. The noodle robots above will burn a lot more fuel than the human variety.
One might argue that humans, if they eat food grown with the help of fossil fuels, might be less efficient than this analysis supposes. Well, do the numbers. People are still better, even if we assume that fossil fuels add a cryptic 10-fold energy budget to human activity. The reason? If we do this, we have to include the enormous costs to manufacture the robots as “embodied” energy.
So, a world with mass robotics would be less sustainable than a human-powered world. Today’s industrial robots are inefficient, but they don’t need unions or medical bills – here dollars substitute for energy lost. But a whole world running with clumsy, inefficient robots? I don’t think we’ll see that any time soon, especially if oil prices continue to rise an alternative energy remains hard to implement (it is).
Robots That Jump will have to be as energy-efficient as humans to compete for their jobs.
This same analysis can apply to Skynet fantasies. James Howard Kunstler’s new book, Too Much Magic, details his interaction with TED – nutcases talking about the Singularity – the moment, sometime in the 2040s, when computers get so smart that one machine can out-think the entire human race. Kunstler is often too far in the other direction – his vision of the future is a “world made by hand” after the oil runs out. But here, I agree with his analysis of TED-driven foolery.
If you’ve never heard of The Singularity, here is a description based on Ray Kurzweil’s vision. Computers keep getting faster, and they get faster faster. Faster faster faster, and by 2040 or so they are smarter than everyone. In the final days, they are doubling in speed every hour, then every minute, then every second. At the moment of singularity, they become infinite.
In a flash, they upload our primitive brain patterns, and then seek to convert the entire universe to (machine) intelligence. Others have commented on the similarity to the Biblical rapture, but Kunstler’s (and my) point is that the energy doesn’t exist to do such a thing. Singularians typically draw an exponential curve, representing computer power, going to infinity in a finite time.
The cool thing that Kunstler asked the TED-Talkers is this:
Where does the power come from?
They’d never even thought of it. It is a classic example of confusing computing with symbols – symbols are “weightless”, whereas embodied symbols in a computer memory use both matter and energy to exist. Unlike a crappy computer game, where you can die and endlessly be restored to life, computer power is real and finite.
Now, unless we assume that there is an unlimited source of energy (presumably discovered by the machines as they become godlike) the Singularity will run out of power long before techno-rapture. Right now, computer circuits are being refined down to wires a few atoms wide, with a countable number of electrons needed to make a logic decision. Soon, we’ll reach a point where a single atom is doing the computing. What then?
Even with “quantum computing”, at some point the Singularity must suck up more and more power. To reach godhood, the computers will need all the energy of the planet, making them Type 1 civilizations on the Kardashev scale. To expand through the stars and down all the cosmic wormholes they’ll need to grab all the energy from the sun, (Type 2 civilization), and ultimately the galaxy (Type 3). This link on Yahoo answers shows that even if we do “black hole computing” or other exotic methods, we will hit limits. Hardly good for an all-seeing, all-knowing god-computer.
There are LOTS of computer problems that can’t be solved – even if every atom in the universe was part of a massive computer. My guess is that someone versed in abstract math could show there are, in fact, an infinite number of problems that can’t be solved by any finite machine.
The point is not whether an algorithm can be created that is more efficient. The point is that there are problems that can’t be solved, even if the entire universe is turned into a computer. So, Singularian ideas are indeed closer to religion (where there is an endless source of magical spirit-energy) than reality.
In fact, most visions of computer-based high tech lack any thought of sustainability. Like the TED guys they ignore energy and often involve perpetual motion. If you look at the plot of the popular Matrix movie series of a decade ago, you’ll note that machines, lacking power, suck energy from humans in pods. These “electricity vampires” apparently solve their sustainability problems with the “energy contained in every human body”. Since there is no sunlight, the put the human race into pod-bottles, and they suck energy from them. They use this energy to survive, and also use it to manufacture artificial food to keep the pod-humans happy. And they’ve got so much energy to burn, they manufacture lighting bolts to make the pod-place look khullll….
Woahhhh….Now granted, this movie was really allegory – in particular, “The Allegory of the Cave” – after Plato. It also borrows heavily, and unconsciously, from “The Prophet’s Journey” in Strauss and Howe “Turnings” theory. But a quick look at this chart shows it is unsustainable.
With no outside power source, the machines will suck energy from humans, lose a large percentage as heat, and only put back a fraction in manufactured food piped into the pods. The system would almost immediately collapse. And, if we add an outside source of energy to take up thermodynamic loss, why bother with the humans in the first place?
But how many tech-freaky, dumb-asses have you heard say “I’m worried we’re in THE MATRIX RIGHT NOW”. Yeech.
The people-pod installations shown in “The Matrix” must have taken vast amounts of energy to construct – and they will never recover their “embodied energy” by sucking 20 watts of so of free energy from an immobile humans.
In short, robotics, machine intelligence, and the like are as yet untouched by any consideration of sustainability. Our visions of the future have a quaint 1950s feature to them – partly because they implicitly assume vast amounts of energy that no amount of thermodynamic waste can affect. One way we will know when Robots That Jump are near is when sustainability appears in the robotic lexicon.