Japanese Robots: On DARPA, SCHAFT, and the Peace Constitution (and bad reporting)
Posted in: Today's Chili
One could reasonably assume Japan’s impressive array of supertech humanoid robots would swarm the inspired-by-Fukushima DARPA Robotics Challenge (DRC), but Tokyo University spin-off SCHAFT Robotics is it. DARPA’s militariness and Japan’s Peace Constitution complicate. Oh, and way-overblown gee-whiz coverage of the DRC isn’t helping.
• • •
A few months back, our own Japan’s Robot Renaissance (Fukushima’s Silver Lining) detailed how a complete unreadiness to respond to the Fukushima component of the 2011 Tohoku Disaster woke up a nation, and the world, to the fact that Japan’s assortment of fantastic humanoid R&D platforms and entertainment robots and robo-buttcheeks, while awesome achievements, were painfully useless for inspection, repair, rescue, or recovery.
It wasn’t until weeks after the initial disaster that American firm iRobot’s PackBot and Warrior robots were sent in to assess; it took months to get a homegrown Japanese robot in there. And that really stings, because we now know a great deal of the damage & pollution was avoidable – if only we could have closed or opened some valves, reconnected a hose, turned some knobs, etc.
Sure, iRobot’s machines were very helpful, and other vehicle-form robots could do a lot of good. Ultimately, however, emergency response experts reached consensus around the notion that, as the majority of humans don’t get around on tank tracks or wheels, when disaster strikes an environment designed for bipedal mammals what we really, really need to safely get in there and get things done is a capable, robotic facsimile.
Of course disaster breeds alarm, and Fukushima put humanoid robotics efforts into competitive overdrive; the silver lining reached all the way across the Pacific.
Hello, I’m the DARPA Robotics Challenge
Okay, DARPA should either be commended or made fun of for sparing almost every expense on graphic design. You be the judge.
Getting to business, know that descriptions of the DRC tend to be either: dry detailed (boring), dorkily detailed (obtuse to laypeople), overly simple (missing the big deal), or the worst – sensationalistically fantastical (the sky is falling oh god oh god killer robots are coming to eat your babies). Hopefully some straightforward sanity to follow – here’s what’s needed to get reasonably hip:
• First Thing About the DRC – Motivator:
Prior to the Tohoku Disaster, certainly Japan, notably the U.S., Korea, and Germany, and many other public and private robotics initiatives around the world were seriously considering the needs and feasibilities, but they were rather casually and quite slowly developing humanoid rescue & recovery robots (ex., prior to the DRC, the U.S. Navy had already begun work on the humanoid Shipboard Autonomous Fire-Fighting Robot (SAFFiR), but, you know, not in a big hurry). There was no specific focus among a broad range of creators, no essential motivation, and no potential for the big, public reward of success.
• Second Thing: A Basic, Bare-Bones DRC Description:
The DRC is an unprecedented two-year contest with cash prizes (though the prestige is arguably worth a lot more) for teams who can make a humanoid robot capable of semi-autonomous disaster recon, rescue, recovery, and repair. If you don’t have your own robot but do have software than can represent, DARPA might give you a robot to prove it.
Ready, GO!
• Third Thing About the DRC – How to Win:
What must be done to win those (relatively few) millions in cash, garner invaluable prestige, and quite likely secure years of lucrative and prestigious robotics contracts around the world? Quoting, the DARPA Robotics challenge aims to:
…invigorate efforts toward developing robots that can operate in rough terrain and austere conditions, using aids (vehicles and hand tools) commonly available in populated areas. Specifically, we want to prove that the following capabilities can be accomplished [by the robot]:
1. Compatibility with environments engineered for humans (even if they are degraded)
2. Ability to use a diverse assortment of tools engineered for humans (from screwdrivers to vehicles)
3. Ability to be supervised by humans who have had little to no robotics training.
…get humanoid robots to successfully demonstrate the following capabilities:
1. Drive a utility vehicle at the site.
2. Travel dismounted across rubble.
3. Remove debris blocking an entryway.
4. Open a door and enter a building.
5. Climb an industrial ladder and traverse an industrial walkway.
6. Use a tool to break through a concrete panel.
7. Locate and close a valve near a leaking pipe.
8. Replace a component such as a cooling pump.
…apply the DARPA Challenge model in order to:
1. Increase the speed of advancements in robotics
2. Grow international cooperation in the field of robotics
3. Attract new innovators to the field
…proceed along a very ambitious timeline:
1. June 2013: Virtual Robotics Challenge (software is running now!)
2. December 2013: DRC Challenge Trials (physical machines)
3. December 2014: DRC Challenge Finals (best of the best, software & machines)
Unquoting.
Among those of us with executive-level robo-dorky proclivities, the DRC is basically one of the most exciting events possible. But the idea of a supertech capitalist competition captures global curiosity and wonder even for those with only a passing interest in robotics.
So it’s underway, and a very international field, including a number of American teams, teams from Spain, Poland, the U.K., Korea, Israel, etc., are now locked in at full-speed. But, oddly, there’s only one team from Japan.* They’re in Track A, which means they’ve got their very own advanced robot and software. But just one team – a small one at that – seems a little… well, it’s Japan, not like it would require reinvention of the wheel: there’s the well-developed ASIMO** and the various HRP robots, as examples, and per the parameters outlined by DARPA, they’re already kinda more than halfway there.
What gives, Japan? Because, as is, this competition could accurately be named or subtitled something like “The What We Really Needed at Fukushima DARPA Robotics Challenge.” More on that in a minute, but first – about that one team:
Team SCHAFT, Tokyo:
Three months ago, the rendering below was pretty much the only publicly available image of the DRC contestant from Tokyo University’s JSK robotics lab spin-off, SCHAFT Robotics:
Even DARPA is still using that image at the DRC homepage, and it doesn’t exactly inspire – there are plenty of teenagers who could render that in an afternoon.
But researchers formerly of a place like Tokyo University are not to be underestimated. Tokyo University is like having the academic disciplines of Harvard, Yale, Princeton, and MIT rolled into a brain trust institution comprised of the highest-level human intelligence Japan has to offer. Not surprisingly, led by CEO Yuto Nakanishi, the small firm of young and ambitious roboticists have really, really brought SCHAFT to life:
There isn’t a whole lot of public info on SCHAFT, but what we do know is that it’s influenced by some 30 years of Tokyo University’s robotics experience, i.e., SCHAFT has a both a serious mechanical pedigree and some very fine-tuned software. Perhaps the the most widely reported feature of this robot is that, within a certain range of motion, it’s limbs can apply more force than a comparably sized human being (Sorry, there are no superstrong-in-general humanoid robotics out there. Yet.). One can get deeper historical details on the SCHAFT team and their university lab’s background, but the world definitely needs more contemporary information on this robot and the motivations of its creators. (Note to Author: You live in Japan, right? Umm, get thee to Tokyo?)
Here’s SCHAFT turning a valve a human can’t handle:
SCHAFT’s considerable advantage in physical strength is possible through a unique cooling system that prevents overheating in its nearly maxed-out electric motors (hence the strength). Another advantage, illustrated below in the image’s translated quote, is the team’s observably high levels of pure, enthusiastic robogeekery – this is a very good thing.
Team member inserting SCHAFT’s coolant; being robo-geeky on TV:
For SCHAFT in motion, the video below includes a brief feature from an NHK documentary on advancing robotics projects around the world. There are some good shots, but the doc is sprinkled with a lot of supposition, and some pretty glaring inaccuracies and generalizations are used to set up unfortunate leaps of logic and just, you know, misstatements. It’s either poorly researched, or very poorly translated, so consider it a nice visual presentation, but when it comes to facts & figures and specific details, definitely not verified or reliable reference material:
(for SCHAFT, jump to 25:25):
So, SCHAFT is cool, highly regarded among other contestants, and well on its way to a good showing at the December 2013 DRC trials. But it’s a curious thing that, with cash prizes and the invaluable prestige of doing well in a wholly unprecedented global robotics challenge, SCHAFT is the only Japanese name in the game. So again, what gives, Japan?
Well, the “D” in DARPA of course stands for Defense. As in United States Department of Defense. As in, humanity’s most massive and far reaching military force like… ever. By far. This doesn’t exactly sit well, and it butts up against a pervasive anti-war sentiment enshrined in modern Japan’s peaceful-by-law society (yep, by law).
The 1947 Postwar “Peace” Constitution: Not So Comfortable With Military-Funded Robots?
Article 9 of Japan’s postwar national constitution is regarded as an explicit prohibition against state-sponsored/perpetrated offensive military activity. So, with a Japan not allowed to build offensive war machines, that has even run into trouble providing tertiary supply line support to allied forces abroad, building robots with cash from the U.S. military is… sticky.
While a point of debate and political grandstanding in Japan, the Peace Constitution has never been amended and it’s unlikely to be anytime soon. And so Japan can defend, but cannot offend, as it were. Obviously this doesn’t prevent private industry from developing machines that might one day make their way into military support roles, though that’s not exactly… approved of.
Japan isn’t the only country to question DARPA’s motivations and express concerns about the DRC leading to some seriously scary Terimator-like murderdeathkill-bots. Last fall, at a conference in Osaka, DARPA’s Gill Pratt responded:
“The DRC is about developing robots that I believe wholeheartedly are completely impractical for military purposes, for offensive military purposes. Will the technology that we come up with find its way into military systems, probably yes. But I guarantee you that if you work on a robot for healthcare, there’s a chance that technology will also find its way into military systems.”
Okay, Japan’s uneasy, but there appears to be a bit of cherry picking with this. After all, Mitsubishi long ago purchased the recipe for American F-15 fighter jets and manufactured them for the Jieitai, the Japanese Self-Defense Forces. Or more subtly, the last two Sony PlayStation consoles probably have the graphics processing capability to guide cruise missiles.
So, maybe, perhaps, probably: it’s the visibility of a military-funded humanoid robotics project vis-à-vis palpable anti-war sentiment that permeates a massively parallel-thinking, group-oriented monoculture. In the form of Honda, robotics pioneers at Kawada Industries, the JSK lab at Tokyo University, AIST, METI, and other public and private robotics developers, Japan has to be aware of how its peaceful-by-law reputation might suffer if it helps build what could be construed as an offensive-capable humanoid warbot.
The final DRC contestants, and especially the winner, are going to be everywhere in the news, and, as is already happening, non tech-focused reporting outlets (and unscrupulous blowhardy loudmouths in general) are going to frantically excrete large quantities of disingenuous, irresponsible, SEO-bating headlines like:
“Meet DARPA’s Killer Android Terminator DeathBot,”
…which could easily metastasize toward:
“Formerly Aggressive Japan has a Killer Robot Soldier – Should We be Afraid?”
So, it’s complicated. And that’s where the story ends. There’s no red bow with which to tie this one off – it’s just complicated, man. Perhaps one will venture to Tokyo, nail down some more SCHAFT details, and discover the identity and motives behind the mysterious Japanese software-only “Team K.”
• • •
Addendum on Weak Robotics Coverage, Media Hype, and Misinformation
There are excellent sources of responsible robotics news out there on the intertubes: IEEE, Gizmag, The Verge, Robohub.org, The Robot Report, Anthrobotic.com, and the URL where you’re currently located. However, outside of Al-Jazeera English and occasionally the BBC, mainline robotics coverage, in the truest sense of the words, produces what is usually half-researched, half-suppositioned, half-assed sensationalism.
They’re far from being alone, but since they published this poster child for unfortunate journalism just a few days ago, today The Guardian gets the blaster: “Darpa Robotics Challenge: the search for the perfect robot soldier.” Karl, this is not good. Karl, is it only about pageviews for you? Karl, do you even want to share any meaningful info? Karl, how long have you been interning over there at the Guardian?
Maybe it can be dismissed as playful journalism, but there’s a huge glaring gigantic wall between playful and irresponsible. Smartassery and pointed, perhaps ironic hyperbole in tech coverage is very, very cool – if, IF it’s qualified and not allowed to fall in love with itself and become a self-sustaining fusion reaction of assclownery for its own sake.
Or, in Karl’s case, hyperbolic scare-mongering to get more views. It doesn’t inform. It doesn’t help. It retards progress and understanding and retards the possibility of developing an informed, nuanced point of view.
As the DARPA Robotics Challenge proceeds, shall we all stop that? KTHX.
• • •
*According to the DRC website, there’s another Japanese Team in Track B (software only), but there is next to zero public information about the group known as “Team K,” and it’s unclear whether or not they’re like, you know, doing anything. (Note to author again: You live in Japan, right?)
**It should be noted that, while unwilling to toss a beefed-up ASIMO into the DRC, Honda is working on their answer to what the DRC will produce. We’ll follow up with some inside info on that later this summer. (Another note to author: Because you live in Japan, right?)
• • •
Reno J. Tibke is the founder and operator of Anthrobotic.com and a contributor at the non-profit Robohub.org.
VIA: SCHAFT (English); DARPA
Images: NHK; SCHAFT
