Robot Bodies Needed Before Robot Minds
Monthly Archives: January 2012
January 14, 2012Posted by on
At the beginning of the year, there’s a lot of optimism. For the last few years, the stock market has been positive, along with the US economy, only to suffer “slowdowns” later in the year. In the area of robotics, we see both optimism and willful delusion at its peak for the year right now.
Jump to CES 2012, and an interview between iRobot and Forbes reporter Elizabeth Woyke.
Well, Ava is a telepresence robot – the kind of remote-controlled metal puppet that robot science has been experimenting with for 30 years, if not more. The biggie is not creating a telepresence robot that looks like an animated broomstick – it’s using an iOS or Android tablet to run the robot’s so-called “brains”. Apparently, iRobot is counting on Google and/or Apple to adapt features designed for cellphones (e.g. face and voice recognition) so that this can be applied in a standards-based way to robots. Great, since iRobot doesn’t have to create custom programs for the same.
Interestingly, Ava isn’t even on the iRobot website – just a trade show bunny.
Certainly, using a tablet for the computer might help in robot adoption. There’s been no real success at creating a “robot OS“, and so the boost to 1980s personal computers provided by DOS and Mac OS has been missing. Tablets are relatively cheap and a huge community exists to program them. The “app” model tends to produce single-purpose programs, often of low enough complexity that a lone programmer can produce quality code. And the app community might help develop “robotic apps” in the mode that has led to an explosion of apps on cellphones and tablets. The potential is similar to Microsoft’s Kinect SDK. What’s not to like?
I suppose to a young Millennial reporter from Forbes who doesn’t know anything about the past or robotics (sorry, the Millennial generation understands even less of the past then previous gens) – Ava must seem like something that “must happen”, “tomorrow today”.
To clarify, there are lots of telepresence robots, both old and new. Many of the groups making them, (including iRobot) have fielded several models starting about a decade ago. The current tech-everywhere boom, in contrast to a generally down economy, has made these robots interesting again:
Here’s one service/teleoperator robot actually doing stuff, from VGO Robotics
And, we cannot leave out this Google tele-operated robot worthy of Big Bang Theory:
And my old friend, Fred Nikgohar, at RoboDynamics:
Boy, singularityhub.com (where I found Fred’s new stuff) has the Techna religion in full swing, witness this quote:
These were the guys warning us about the coming “End Times” 1000 years ago, if reincarnation is a reality…
Another “blast from the early 2000s” – Anybot
This in itself is hardly “innovation”, except that cheap, mass-produced computers are being used to operate the robots. Tele-operated robots with enought height to show a screen “face” to humans is an idea many years in the making. They were being being tested in “trial” runs in 2002. iRobot’s Ada looks behind the curve, rather than leading it, though I saw no mention of the tele-operated robot they released right after their floor-sweeper.
Now, onto the claims and willful delusion of the reporter, in the main article on Forbes.
- Ava is “intelligent” because “she” (um, where are the genitals) has iOS or Android apps installed. That’s what the article says. This is crazy. Apps are not powerful enough to produce meaningful intelligence. IBM’s Watson needed several-thousand high-performance CPUs to produce one feature of human memory – deducing a question from an answer. A low-power, mobile device running a robot can’t possibly produce intelligence of the “Ai” variety.Why does the author say this? One possiblity is that, in the mind of the tech-faithful, robots are becoming intelligent, and will soon be smarter than us, end times, skynet, upload, immortal… (gag).Or, for this author, intelligence is the same thing as running a mobile app. It is as if intelligence and “good app” are the same. Such a statement tends to validate those old hippies out there who complain that Millennials lack the ability to question their own preconceived notions. A bit of reflection should make it obvious how stupid this is. If that’s the case, I recommend she read: “You are not a Gadget” (though it’s too long to text)http://www.jaronlanier.com/gadgetwebresources.htmlApps are simply not intelligent. The only way that would happen is if the robo-app Ajaxed its way into a monster server cloud, which is suppose is possible someday. The reporter confuses utility with intelligence. She maps gadget-ness to having a mind.
- Ada is special because of her height (nope, that has been done for about 30 years).
- Ada is special because she has emotions.This was irritating enough to motivate me to write this entire blog. It is another example of the bizzare vitalism that has resurfaced in a supposedly secular, scientific culture. Whatever Ada does to flash “her” lights is not emotion. The mammalian emotional brain is extremely unlikely to resemble iOS in any way. There are unlikely to be discrete programs creating emotion, software, discrete states, variables, etc. Ada most likely has some programs sampling recent sensory history, and then applying some state engine rules to jump to an “emotional” state.To let Mark Chiappetta blather on about this “female” robot’s emotions (girls are more emotional, of course and flash pink) is the sign of a nonprofessional journalist.“…Ava’s color-altering moods don’t flare up often…” – Elizabeth Woyke, Forbes OnlineApparently, asking Mark if Ada’s emotions were real, or if tested they would match, or at least mimic, animal emotion would have been like farting – too impolite. It’ like someone interviewing a priest and asking about “the meaning” of their faith, carefully ignoring hard questions whether a saint really did appear on a slice of toast.In fact, Ada is a big metal puppet. There’s much that is cool about that – it is what Aristotle wrote about when he imagined replacing human slaves with machines 2500 years ago:“…There is only one condition in which we can imagine managers not needing subordinates, and masters not needing slaves. This condition would be that each (inanimate) instrument could do its own work, at the word of command or by intelligent anticipation, like the statues of Daedalus or the tripods made by Hephaestus, of which Homer relates that
‘Of their own motion they entered the conclave of Gods on Olympus…’
as if a shuttle should weave of itself, and a plectrum should do its own harp playing.” – Aristotle, Politics
But, when I watch a puppet perform I don’t think that its movements make it emotional, unless I’m a gullible rustic like those visiting temples 2000 years ago where Heron of Alexandria put his robots. I bet the robot which hauled his “automatic play” to the stage looked a lot like the broomstick robots above.
I might talk to a small child about the emotions of Kermit the frog, but kids more than a few years old know that the frog puppet doesn’t feel anything – it is “tele-operating” emotions created by its manipulator.
And, I’m not knocking priests. There are plenty of religious people who enjoy debating their beliefs. A typical Jesuit these days (though not in 1500 A.D.) would be happy if a chance for debate about the reality of God or soul came up – they’ve thought it through and through and can discuss it rationally, if not always convincingly.
However, the new tech-religious faithful are more afraid of insulting their god, or their high poo-bahs at trade shows by asking if robots like Ada actually have emotions. It’s the new machine-vitalism.
This article is one of hundreds on the web where the reporter blindly accepts any crazy thing said about arobot, since they apply only their irrational belief in Techna, and not their knowledge of robotic science. iRobot may have created a useful robot, and I’m not knocking that. I’m attacking too-easy belief in “the ghost in the machine”.
Now, if you say we will get more funding for robotics if we pretend the metal puppets are alive, you are exactly like a priest operating one of Heron’s “temple tricks” who lies to keep the unwashed believing in their religion.
Hey, my programming skills are modest, but I could write code for a room-dimmer that used ultrasound and a microprocessor to detect a person entering a room, and flashed pink lights if they “left too soon”. Is this anger? Have I created emotion? The only way Ada can have emotions is if we accept that said room-dimmer can be “happy” or “sad”.
We could apply a Turing test (essentially 1950s Behaviorism applied to machines) and say that if Ada flashes the right lights at the right times, she has an emotion. This is acceptable only if actions are the only reality, and inner states (mechanism) is irrelevant or don’t even exist.
But if you believe that, emotion in humans isn’t any different from flashing lights when you bump into something. And, if that’s the case, the scream of a dying human is not different from the reboot bell on a PC. I suspect that there’s a lot more going on in the human versus the machine.
But the Forbes article, supposedly produced for critical-thinking adults, has a Santa Claus like acceptance of the intelligent and emotional Ada, along with an implied future rise of her brethren. Ada may be a useful metal puppet, but the idea of her iPad brain having a “mad app” is crap. I doubt if Ada Byron herself, even in the most kooky of her drug-induced dreams, would have accepted this idea.
Here’s my advice to everyone who writes such faith-based machine flapdoodle. Stop worrying about the mind of a robot. Stop acting like a puppet really has feelings. Oddly enough, the same reporter did an article on RoMeLa, the Robotics & Mechanism Laboratory in the forefront 0f creating Robots That Jump.
Unlike the babbling about the “mind” of a robot rife on tech-blogs and in Singularian bibles, robots that move confidently in the environment don’t require speculating about a ghost in the machine – they either do or they don’t work. My feeling (unlike Ada, who doesn’t have any), is that Robots that Jump are the real cool.
January 9, 2012Posted by on
The never-ending drumbeat for the rise of the robots imagines them super-strong, super-fast, and able to recover from massive injury and keep on going – sort of like mechanical zombies. In reality, robots today can’t jump because they are so fragile. This must explain the puzzlement techno-freaks have over why Japan, with its huge robot efforts, couldn’t field any of these “ready to take over” models in the Fukushima nuclear disaster. In fact, the best that they could do were tethered tele-operated machines.
“According to experts, the biggest reason Japanese robots such as Honda’s Asimo were not used early on was their vulnerability to high radiation levels, which could easily damage their integrated circuits.
The domestic robot industry, in fact, had stopped working on ways to shield robots from extreme radiation around 10 years before the Fukushima crisis, and manufacturers and institutes were caught completely off guard, experts said.”
This story shows how little robotics has actually advanced in decades, in terms of real-world useful devices. The robots that actually went in site were US company iRobot Packbots.
These robots are not much different (though cheaper and sporting better sensors) than bomb-squad robots produced 20 or more years ago. In fact, I remember robotic, tele-operated arms being used in nuclear reactors in the late 1950s. They, and Packbot were built from a “body” perspective – a Robot That Jumps – though they were, and are, dumb as a stump.
The advanced Japanese “domestics” could not enter the reactor area for the same reasons that humans couldn’t – radiation. As microprocessors have become smaller and more complex, they have become ever-more sensitive to radiation. An Asimo entering the reactor area would have instantly fried its circuits. This combines with the sensitivity of robots to Electro-magnetic pulses, or EMPs, to paint a picture of a fragile lab experiment.
So, it’s not a surprise that “radiation hardened” robots like the PackBot are really tele-operated, with minimal autonomy. It is just too hard to make a complex brain in silicon that can also ride on a mobile robot body. The way we’re making more complex brains makes them more and more delicate.
NASA, which operates robots in environments with high levels of radiation, handles the problem simply – use old microprocessors. The Mars MER rovers Spirit and Opportunity used scaled-up versions of CPUs used in Macintosh computers from the early 1990s. The clock speed is 33MHz, about 100 times slower than a fast desktop today. My understanding is that the chip is radiation-hardened by making it much larger (with thicker wires) than the standard chip (a full discussion of radiation-hardening at http://en.wikipedia.org/wiki/Radiation_hardening). The take-home is that you can’t just put a shield around a modern chip – you have to redesign it, and go back into the past.
A modern 2012 Macintosh, or a comparable Windows system, would fry in the Martian radiation environment – and it’s not really that bad compared to Earth. As it is, the MER rovers have had several instances of memory corruption, despite radiation-hardened memory – an errant cosmic ray required a scary reboot of the system.
We could see a similar idea at work in the Space Shuttle, which had the luxury of being below Earth’s radiation belts and relatively protected. For many years, the Shuttles continued to use 386 processors (pre windows 95 stuff). The latest and greatest, at least the computers you would would get at Best Buy, and whose microprocessors are the same as most robotic efforts, were just to fragile. While radiation was less of an issue, fragility probably was. It would be very interesting to find out if the laptops the astronauts bring up are subject to problems, even in low Earth orbit.
Why are we at this stage? Part of the problem is the very technology enabling advanced robotics – silicon-based processors. If you make the wires small, they’re vulnerable. To reach the complexity required for a robotic brain, the chips become fragile. As we iterate through the final shrinking of this technology to wires a few atoms wide, we will end up with exquisitely sensitive chips. The will be powerful, but will require massive shielding to work in any real-world environment.
So, unlike the dorks at Slashdot think, the robot race isn’t about to take over yet – instead of a shotgun through the head like a zombie, just get a radiation source and – zap! Fascinating that Asimov had his positronic robots vulnerable in the same way – in more than one story, a gamma ray blast was able to destroy the robot’s brain. Strange that we’ve forgotten this (or don’t want to remember) and movies are made with robots that can’t be destroyed.
I suspect that robot’s still won’t be ready for the next natural disaster – at present, the human body, or simple teleoperated devices, are the way to go. However, it is interesting that the Japanese are going to try automated farm machinery in the area damaged by the tidal wave.
The best irony – instead of showing us what an automated farm machine might look like, we get a picture of the Asimo with the morning $5 double extra frapmeister chinuscope coffee, as if a robot that can’t even go outside is going to farm. This is the silly gap between robot fantasy and reality.