Thomas Aquinas and the DARPA Robot Competition

Korea's Team Kaist's DRC-Hubo humanoid robot walked — and drove — through DARPA's 2015 robot competition: one of three to complete every task on the course without falling over.

Meanwhile, Toshiba's (somewhat) lifelike ChihiraAico robot demonstrated 'her' voice in a Las Vegas trade show.

1. DARPA Robotics Challenge 2015
   • Hope and Common Sense
2. "Like Drunks Trying to Pass a Sobriety Test"
   • Whoops
3. Welcome to Uncanny Valley
   • Passing a Turing Test

---

Simple Tasks: for a Human

(From DARPA, via BBC News, used w/o permission.)
("Team Kaist's winning robot - DRC-Hubo - was able to drive a car..."
(BBC News))


(From DARPA, via BBC News, used w/o permission.)
("... and then get out of it without falling over"
(BBC News))

When I first glanced at that photo of DRC-Hubo driving, it looked like the robot was adjusting a sun visor.

These days, a robot driving a vehicle isn't all that remarkable. What is newsworthy is that the same robot got out of the car without falling over, finishing every one of the contest's tasks:

• Driving a car
• Getting out of the car
• Opening and walking through a door
• Opening a valve
• Using a drill to cut a hole drawn onto a wall
• Crossing a debris-filled terrain - either by clearing a path for itself or walking over the rubble
• Climbing up steps
• A mystery event, that will only be disclosed to the teams on Friday
  (BBC News)

Most adult humans could complete the list easily: although using a drill takes practice — talk to a high school shop instructor if you're doubtful about that.

The point of this contest is that there are places humans should avoid: like a broken nuclear reactor. Remember Fukushima Daiichi and Chernobyl?

On a smaller scale, robotic firefighters could save lives by operating in areas that would kill a human. We have comparatively narrow tolerances for temperature and atmospheric composition, and hazmat suits have their limits.

Meet the Robots

(From Team Walk-Man, used w/o permission.)
("The robots will compete at Pomona, California, which is a short drive from Los Angeles"
(BBC News))

That's Walk-Man, from the Istituto Italiano di Tecnologia/Italian Institute of Technology and University of Pisa in Italy, part of the European WALKMAN project. (www.walk-man.eu)

DARPA Robotics Challenge Finals listed the 25 teams who qualified for the finals. Some, like Team Aero and Team Grit, had robots with four legs.

Looking at a short 'blooper reel,' my guess is that most disaster-response robots will be shaped like us. Difficult as balancing on two legs is, it's still a good way to get over a wide variety of uneven and shifting terrain.


(From Google, via BBC News, used w/o permission.)
("Google's Schaft robot beat 15 other teams that took part in the last round of the challenge"
(BBC News))

More:

• "Disaster droids face off in the Darpa Robotics Challenge"
  Leo Kelion, BBC News (June 4, 2015)

Table Tennis, John Henry, and Being Human

Opening doors and walking over rubble is one thing: but can robots perform that pinnacle of human performance, ping pong?

Turns out the answer is "yes." Zhejiang University's Kong plays a pretty good game of table tennis. Apparently the best record for a two-robot game in 2011 was 176 strokes. That's not world-championship performance: yet.

I've talked about robots, John Henry, and coping with change before. Often. I'm quite sure that robots can do some jobs better than humans. But I'm also quite sure that being human is something we'll still do best. (December 5, 2014; August 15, 2014; July 11, 2014)

More:

• "Balance Motion Generation for a Humanoid Robot Playing Table Tennis"
  Yichao Sun, Rong Xiong, Qiuguo Zhu, Jun Wu, Jian Chu; State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou, China; for the 2011 IEEE-RAS International Conference on Humanoid Robots, Bled, Slovenia (October 26-28, 2011)

Using Tools Wisely and Well

St. Thomas Aquinas didn't discuss the theological ramifications of artificial intelligence: mainly because it wasn't being invented in his day.

He did, however, address 'dangerous' technology. This excerpt is on the long side, but I think it's worth reading:

"In the words of Augustine (Super. Gen. contr. Manich. i): 'If an unskilled person enters the workshop of an artificer he sees in it many appliances of which he does not understand the use, and which, if he is a foolish fellow, he considers unnecessary. Moreover, should he carelessly fall into the fire, or wound himself with a sharp-edged tool, he is under the impression that many of the things there are hurtful; whereas the craftsman, knowing their use, laughs at his folly. And thus some people presume to find fault with many things in this world, through not seeing the reasons for their existence. For though not required for the furnishing of our house, these things are necessary for the perfection of the universe.' And, since man before he sinned would have used the things of this world conformably to the order designed, poisonous animals would not have injured him."
("The Summa Theologica," First Part, Question 72; St. Thomas Aquinas [emphasis mine])

---

"...Now action is properly ascribed, not to the instrument, but to the principal agent, as building is ascribed to the builder, not to his tools. Hence it is evident that use is, properly speaking, an act of the will...."
("The Summa Theologica," First Part of the Second Part, Question 16, Article 1; St. Thomas Aquinas)
(translated by Fathers of the English Dominican Province (Benziger Bros. edition, 1947))

St. Thomas Aquinas was quoting St. Augustine of Hippo. Falling into a fire or getting cut isn't a good idea: but if a dummkopf wanders into a workshop and gets hurt — it isn't the fire's fault, or the sharp-edged tool.

Maybe the workshop's owner should have kept the door closed, or paid more attention to the hapless visitor's actions.

Although today's robots are remarkable machines, they are — machines, tools. Whether we use them to help or hurt other people is up to us.

More:

• Catechism of the Catholic Church, 2293-2294
• "Gaudium et Spes," 5, 15, 20, 33, 54, 56, 62, 86d
  Pope B. Paul VI (December 7, 1965)

---

1. DARPA Robotics Challenge 2015

(From Getty Images, via BBC News, used w/o permission.)
("Team Kaist's DRC-Hubo robot completed the course in less than 45 minutes"
(BBC News))

"South Koreans win Darpa robotics challenge"
Regan Morris, BBC News (June 7, 2015)

"A South Korean robotics team has won the Darpa Robotics Challenge.

"The contest is a battle of robots on an obstacle course meant to simulate conditions similar to the 2011 Fukushima nuclear plant disaster.

"Team Kaist's DRC-Hubo humanoid robot defeated 22 others to win the top $2m prize from the US Department of Defense's Darpa research unit.

"The robots had an hour to complete a series of tasks, such as a driving a car and walking up steps...."

DARPA stands for Defense Advanced Research Projects Agency. It's the U.S. Department of Defense agency responsible for developing emerging technologies — with military uses. DARPA's landing page header reads "Creating Breakthrough Technologies For National Security."

An adolescence spent in the '60s gave me a particular knee-jerk response to "military." But since a considerable fraction of folks who earned my respect were in the military, I had to think about my attitude.

That's not quite right. I don't "have to" think, I have free will like everyone else. (Catechism of the Catholic Church, 1730)

But I'm bothered when things don't make sense, so I took a hard look at my conflicting attitudes. I learned about the just war doctrine, became a Catholic, and that's another topic. Topics. (April 19, 2015; August 24, 2014; September 11, 2011)

Getting back to the DARPA Robotics Challenge (DRC), the stated goal seems reasonable:

"The DRC is a competition of robot systems and software teams vying to develop robots capable of assisting humans in responding to natural and man-made disasters....

"...Technologies resulting from the DRC will transform the field of robotics and catapult forward development of robots featuring task-level autonomy that can operate in the hazardous, degraded conditions common in disaster zones."
(DARPA Robotics Challenge 2015)

I'm pretty sure that robots able to operate in a broken nuclear power plant could operate in a war zone, too. I'd be surprised if DARPA did something with no relevance to national security.

Hope and Common Sense

My father remembers his family's first kerosene lamp, I remember when color television was new, and one of my kids helped my son-in-law program an industrial robot. This isn't an ideal era for folks who dislike change. (May 15, 2015; March 27, 2015)

I'm pretty sure that some high schoolers have decided to try living in the past. My guess is that more are on the same page as five who won a DRC-associated video contest, called Robots4Us.

Their videos reflect a refreshing mix of hope and common sense:

"...optimistic expectations that a maturing robotics technology will improve the human condition in many ways, from helping farmers feed an ever-growing population to enabling older workers to stay on the job longer. At the same time, these roboticists-in-the-making expressed concern that hackers, faulty robots, and even robots with their own new kind of intelligence could pose threats to society that would have to be managed...."
("Robotics Challenge Workshop: From Better Robots to Better Futures," DARPA (June 8, 2015))

---

2. "Like Drunks Trying to Pass a Sobriety Test"

(From BBC News, used w/o permission.)
("Team Tartan Rescue's Chimp was one of only three robots that managed to complete all the tasks"
(BBC News))

"Falls of the robots: Disaster droids struggle to stay upright"
Regan Morris, BBC News (June 8, 2015)

"It's a scenario straight out of a Hollywood movie. A disaster too dangerous for humans to enter, so a robot must drive alone into the area, and do the dirty work of inspecting, repairing and clearing the site.

"That was the scenario at the Darpa Robotics Challenge last weekend at the Los Angeles County Fairgrounds.

"The muscly, humanoid robots looked like Terminators, but they moved like drunks trying to pass a sobriety test. We were 35 miles from the heart of Hollywood, but a world away in terms of robot capabilities.

" 'It's like watching paint dry,' remarked Brad Tousley, director of tactical technology at Darpa - the US Defense Department's research unit - as he watched the excruciatingly slow robots attempt to perform eight simple tasks in an hour...."

First, about robots in movies: Terminator 3: Rise of the Machines reminded me a bit of Colossus: The Forbin Project, except that Skynet doesn't have a central processor.

I appreciated Terminator 2: Judgment Day's nod to plausibility — we learn that Skynet's tech is based on the damaged right arm and CPU of the first movie's Terminator, which was built by Skynet.

I've enjoyed some stories involving a time loop, like the device in Babylon 5 that's brought to the space station after being stored by Valen — who took it from the space station and traveled back in time, so it could be brought to the space station.

Getting back to killer robots, Skynet, and the Roomba revolution that wasn't, I think we're a very long way from AI like Colossus, or robots like C-3PO. (August 22, 2014)

Whoops

(From Getty Images, via BBC News, used w/o permission.)
("The Running Man robot tumbled while staging a victory dance"
(BBC News))

"...Perhaps the biggest showman of the contest was second place winner, IHMC Robotics from Florida, which took home a $1m prize.

"Its robot, nicknamed Running Man, was so confident on its second day finish that when it completed the final task of climbing stairs onto a podium, it thrust its humanoid arms in victory, did a little dance and then tripped and collapsed...."
(Regan Morris, BBC News)

IHMC stands for Institute for Human & Machine Cognition — but I'm pretty sure that describing Running Man as "so confident ... that ... it thrust its humanoid arms in victory" is an example of anthropomorphism.

Folks who actually work with robots have, I think, a better grasp of the technology's potential: and limitations.

"...Humans 'get most of our ideas about robotics from science fiction. (Today) we want to show a little bit of science fact,' Gill Pratt, director of the event, told the Associated Press.

"While Hollywood has portrayed robots as agile and self-thinking, the machines are only now beginning to make strides toward that ideal, and the competition was designed to push the research forward, organizers said...."
("IHMC's Running Man captures 2nd at DARPA robotics competition," Florida Institute for Human & Machine Cognition (June 8, 2015))

These robots are shaped — very approximately — like humans, which probably explains the pathos I saw in the picture of the fallen Thor-RD valiantly clinging to a drill, after trying to use the simpler machine.


(From DARPA, via BBC News, used w/o permission.)
("Thor-RD - a robot created by US-based roboticists - managed to turn a valve... but fell over when it was supposed to drill a hole in a wall. It only completed three of the eight tasks"
(BBC News))

Then there's HRP2+ - - -


(From DARPA, via BBC News, used w/o permission.)
("Team Aist-Nedo's HRP2+ robot was able to open a door... but failed to stay upright while trying to walk over rubble"
(BBC News))

I quoted XKCD's "Robot Apocalypse" a few weeks back. I really don't think we need to worry about a robot revolution. Not in the immediate future. (May 15, 2015)

On the other hand, now would be a good time to start getting used to the idea that robots can make themselves useful.

I think Carnegie Mellon University's Clark Haynes is right: Japanese and Korean society is much more likely to see working with robots as opportunities, not threats.

"...Many of the roboticists say they think there is a cultural difference in how societies treat and relate to robots and that Westerners are not as open to robots as Japanese and Korean people.

"Clark Haynes, of Carnegie Mellon University's team who built Chimp, said the school is more like Japan than the rest of the United States because it embraces a future where robots can improve our lives.

" 'The challenge is a collaboration between humans operating the robot and the robot itself,' he said.

"It's not like these are robots with self-intent. The robots are just tools, it's just a better hammer for a unique task.'..."
(Regan Morris, BBC News)

The next item is an op-ed about workplace productivity, and why useless jobs don't make sense. I picked it because it included one of the best photos I've seen of Toshiba's robotic receptionist.

---

3. Welcome to Uncanny Valley

(From The Asahi Shimbun, via Bloomberg Views, used w/o permission.)
("One of them is actually alive."
(Bloomberg Views))

"Even a Robot Shouldn't Do These Jobs"
Noah Smith, Bloomberg View (June 7, 2015)

"Americans are worried about robots taking their jobs. Japanese people should be salivating at the prospect.

"Japan continues to power ahead in the field of robotics. Although Japanese companies have lost their dominance in fields such as consumer electronics and semiconductors, in robots they remain supreme. Japanese companies hold 50 percent of the global market for manufacturing robots and 90 percent of the market for robot parts.

"In fact, Japanese companies' enthusiasm for robots goes way beyond industrial automation. Corporate and government leaders are fond of declaring that as the country ages and labor becomes scarce, robots will take over many jobs now held by humans, such as elder care.

"But there's something even better than robots that could replace large numbers of Japan's human jobs: nothing.

"Japan is a country famous for its low white-collar productivity; this is borne out by the statistics. Some of that comes from the reluctance by tradition-minded companies to adopt modern workplace technologies -- there are still companies using fax machines or copying electronic documents onto paper. Some of it is from outdated management practices...."

About that photo: the one wearing traditional Japanese clothing is a Toshiba ChihiraAico communication robot. Toshiba says a ChihiraAico's primary function will be communicating with humans at information and receptionist desks.

My hat's off to Toshiba engineers, that's a very impressive robot.

As a Toshiba video shows, ChihiraAico can express emotions, and use sign language. A more recent video shows the robot speaking at the CES 2015 trade show in Las Vegas.

The ChihiraAico performing at the trade show effectively describes the basic features of 'her' design — together with a brief discussion of Toshiba's goals.

Another article/op-ed, by Mary-Ann Russon in International Business Times, includes a shorter video: also taken at the trade show, with some voiceover commentary.

I'll agree that Toshiba's ChihiraAico is "lifelike." However, although this communication robot is a remarkable step toward practical robot-human social interaction: once ChihiraAico starts moving, the robot could not be mistaken for a human.

The 43 pneumatic actuators work just fine: but 'her' movements, although precisely and effectively mimicking gestures, are clearly those of a sophisticated robot.

I think 'social robots' will either have to be a whole lot better at acting like humans, or look less like us. Maybe both. My guess is that robots like Cynthia Breazeal's Jibo, that have social skills but don't look at all human, will be more popular — at least in Western countries.

ChihiraAico's a bit too far into Masahiro More's uncanny valley for comfort. My opinion.

Passing a Turing Test

My guess is that the biggest technical challenge will be developing an AI for ChihiraAico and similar robots.

Natural language processing, AI that understands human languages like English or Mandarin, has come a long way since the 1950s. That said, chatterbots like ELIZAneth a rogerian psychoterapist are still far from passing a Turing test.

I'll grant that Eugene Goostman's performance at the 2014University of Reading competition, fooling a third of the judges, was impressive. However —

Here's a conversation I had with ELIZAneth, another chatterbot, a few days back:

• ELIZA: Hello, darling what's up today?
• Me: Hello
• ELIZA: HELLO! WHAT DO YOU WANT TO TALK ABOUT?
• Me: I'm curious. What is a rogerian psychoterapist?
• ELIZA: Do you enjoy being CURIOUS ?
• Me: Being curious is my default state. And yes, I enjoy being curious.
• ELIZA: YOUR DEFAULT STATE .
• Me: Yes default state: a condition which is normal for me.
• ELIZA: I'M NOT SURE I KNOW WHAT YOU'RE TALKING ABOUT.
• Me: I was trying to ask what a rogerian psychoterapist is.
• ELIZA: I SEE.
• Me: What is a rogerian psychoterapist?
• ELIZA: I'M NOT SURE I KNOW WHAT YOU'RE TALKING ABOUT.

Come to think of it, my conversations with humans have occasionally been that satisfactory: particularly when I didn't restrain my penchant for sesquipedalian pleonasm.

The point isn't that having an intelligent conversation with an AI is impossible: just that it's nowhere near as easy as folks thought in the 1950s.

Finally, a quick look at the neurological system chatterbot programmers are trying to emulate.

We've been learning more about how humans process language, and are tracing some of the neural circuitry we use. I've talked about rhesus monkeys, adult humans, and neuroplasticity, before. (March 27, 2015)

I gather that most American kids start using simple phrases and have something close to a 50-word vocabulary by age two. On average. (Mayo Clinic)

There's apparently a growing awareness that everybody isn't average — and that this is okay. Outfits like Mayo Clinic carefully say "your child might" when describing typical developmental milestones.

Even so, my guess is that new parents and relatives will be jittery: many of them, anyway.

My parents told me that I didn't speak at all until long after the usual developmental milestones had sailed past. Some relatives were in a tizzy: but my folks said they figured I'd speak when I had something to say.

They were right. At some point before kindergarten age, I started speaking — in full sentences. The trick since then has been to keep me from talking, and that's yet another topic.

More about robots:

• "A Robotic Tentacle, and Disney's Baymax"
  (May 22, 2015)
  Particularly
  • Frankenstein's Monster, C-3PO - - -
  • - - - and Ultron
  • Artificial Intelligence: Still a Challenge
• "Fire Ant Engineering and Bungee Nerves"
  (May 15, 2015)
  Particularly
  • Stone Tools, Bessemer Steel, and Robots
• "The Thumb-Brain Connection, and DIY Robots"
  (March 27, 2015)
  Particularly
  • Brains and Assumptions
  • God, Clay, and Accepting Arithmetic
  • Beware the Robots! — or — Rotwang's Maschinenmensch, H.A.L., and All That
• "Build Your Own Robot Swarm — or — Angular Automatons and Cuckoo Clocks"
  (August 22, 2014)
  Particularly
  • The Digesting Duck that Didn't, Cuckoo Clocks, and Pygmalion
  • Victor Frankenstein, The Phantom Creeps, and All That
  • Automatons: Not Entirely Lifelike
• "Neurosynaptic Cores and Retinal Implants: Getting a Grip About Tech"
  (August 15, 2014)
  Particularly
  • "Metropolis," Tsukumogami, and the Roomba Revolution that Wasn't
  • IBM's TrueNorth Chip and Cockroach Brains