"indecision may or may not be my problem."
that's it. my thesis is printed and signed by my advisor. still hunting for some signatures, but there's no more work to be done, and no changes to be made. i've uploaded the final pdf for your reading pleasure.
please don't point out any typos in the final version. at least not for a while. however, i might re-edit it for a celebratory binding ceremony held later this fall.
compared to the thesis defense, which felt great, this is much more of an anti-climax, not very special. just another night, another printing job. when all is said and done, it's just a pdf, and not a very pretty one to boot. i don't think i changed the face of science. at least i gave it a fair shot.
somehow this pdf doesn't capture the experience quite well.
in accidental sync with the cuban revolutionary movement day, and after four years of work, my thesis defense is scheduled for tomorrow at 3:30pm in the bartos theater of the media lab (lower level of m.i.t building e15). here comes a date that for years was located "somewhere" in the muddy future, with often a day on which it didn't even seem possible.
for those out of town but who still want to watch, there will be a live webcast (the link becomes active only at the time of the event). 3:30pm boston time is 21:30h in europe and 22:30h in israel.
i'm still negotiating the possibility to have questions come in over skype at the q&a session. email me for my skype id, if you're interested in that.
other than that - see you on the other side.
just came back from the 2nd human-robot interaction conference, and an mit student who works in gehry's extravagant stata center, home of m.i.t's computer science and a.i lab, made a curious comment to me.
while i wondered whether it was a nightmare to run robots in the building's twisted architecture, she said that actually it's almost as if the building was made with robots in mind, because no two views of it look the same, which is great for robot navigation. she is obviously right. robots have limited sensing capabilities and there's nothing more confusing for the little critters than repetitive hallways and identical doorways.
no such problem at the stata center. robots could easily do single-camera memory-based navigation in that beast of crooked lines and tilted walls. the glass walls, though, she admitted - were a bummer for the laser range-finders.
honestly, i doubt that gehry thought about any of that, though.
ok - this seems pretty fucked up. a friend just rang me and called my attention to the following g.m. superbowl ad:
a fake rating notice? a mock movie trailer? a sad robot all alone? and 'all by myself' as the backing music, kicking into the chorus just when the corny movie titles are over?
sort of reminds me too much of a little something i made back in 2004 with our own robot.
coincidence? you be the judge. i'd like to talk to the person who came up with the g.m. idea.
later, i discovered the full version:
update: this post has been updated since it was first posted - i've added embedded video for easier viewing. you can still follow the links, though, if you prefer.
some impressive new videos of recent robot locomotion work:
robots seem to also be the hot news item of 2007, more than ever before. harper's title page is dedicated to the coming robot army, and bill gates claims that robotics is the next big thing. the only sad thing is that microsoft has decided to meddle with the field.
working as i am with robots, people regularly ask me to interpret their hopes and fears about our metallic kins. an oft-recurring question is 'will robots take over the world?'.
i think the following video pretty much answers that question
and if the operators look surprised to you - you might want to take note that they had the mobile barrier in place well ahead of time.
by god, i know exactly how they feel.
p.s. - wouldn't it be cool if the robot had some kind of sensor to detect a situation like this, which would start a different sound sequence, something like 'damn, this always happens to me. shit, that hurts!'. not that i know what he's even saying in japanese.
Theater actors have been staging artificial intelligence for centuries. If one shares the view that intelligence manifests in behavior, one must wonder what lessons the AI community can draw from a practice that is historically concerned with the infusion of artificial behavior into such vessels as body and text.
this work is based on the longer Four Lessons paper.
beck's music video for 'hell yes' (incidentally one of my favorite songs from the new album) shows off 4 dancing sony qrios. the combination of japanese dancing robots with mtv choreography, cinematography, and editing creates a product that - to the best of my knowledge - is unprecendented.
thanks to the annoying flash navigation, there is no direct link. you'll have to go to beck.com and navigate to 'Video' and then to 'Hell Yes'.
note: the video i mean is not the one labeled 'One Shot', although that one is also cute.
our lab's online robot teaching study is back, new and improved.
if you haven't tried it out yet, here's the deal: we're researching the way people interact with robotic agents (and more generally "learning agents"), and as part of that we're running an online study where you get to teach a robot game character how to bake a cake.
everyone with a java-enabled browser can try it out, so you should, too!
once you're done, you're asked to fill out a short survey. and this is not 'our check-in lines are short' short, it's really short. the whole thing (game+survey) shouldn't take more than 15-20 minutes. and you'll get to advance science from your bedroom (just think how much longer a ph.d takes), as well as a chance to win $100 on amazon.
if you don't have the 15-20 minutes now, come back when you do, because we can only use the data of people who complete the whole process.
if you're interested, i'm pretty sure we can send you the research results once we've finished analyzing and publishing them. the first pilot we ran already showed some pretty interesting stuff.
the woman on the other line is seriously disturbed by her cleaning robot 'looking at her funny', and 'conspiring against her'. also she has apparently been abducted by aliens which caused her to get evicted. all that while she has to tend to 6 or 7 children (she can't remember how many), which is damn hard 'while you're up there in the spaceship'.
not everything is intelligable, but this is what i managed to transcribe so far.
from what i understand, she was trying to contact iRobot, the m.i.t spinoff that brought you the 'roomba' vaccum cleaning robot, and is also busy designing robots for the u.s army.
what do white furry felines, an ikea-furnished singles apartment, fake bathroom tiles, cartoonish slides, and french accordeon music have in common?
we're on the road to electric sheep no doubt.
and not to stir up the whole canine-feline debate, but of course it's easier to model a cat robot than a dog one. cat's are stupid.
possibly of very limited interest to most readers, i have put my thoughts from a month of reading about acting onto paper. h.r.i: four lessons from acting method is an informal paper describing some points of inspiration that robot designers might take from the way actors prepare and work. it is free-flowing, speculative, and forward-thinking. but after three days of tinkering around, i think it's in a distributable form, while admittedly not nearly finished or perfect.
Robot design should consider tearing down the implicit barrier between the motor system and the behavior system and think afresh about a combined architecture where both are sides of the same behavior. Motion should not only influence thinking (as it sometimes, but rarely, does), but should be the decision process.
from "metal performance: humanizing robots, returning to nature, and camping about", by steve dixon, The Drama Review 48, 4 (T184), Winter 2004.
Although robots may not yet be self-aware, they are quintessentially "self-conscious" entities, calculating and computing their every move.
But more than anything I couldn't help but notice how much fun the crew must have had making the movie. It is chock-full of just random jokes and ideas, some so psychedelic that they make no sense in the script except that someone was like "wouldn't it be cool if" and then the director and producers were just 'hey - go ahead and do that, if our employees have fun (and not just in the mission statement), the movie will be fun'.
This as opposed to pixar movies (including the great 'incredibles') that really give off a sense of hard work and a tight-run ship more than anything else.
the details are: thursday feb.3 4:30pm, pound hall 102.
i wonder if a second screening will be as impressive, the discussion should be fun nonetheless. chatting about how robots are going to change our lives has been a favorite pastime since homo sapiens started walking the face of the earth.
people tell me to wear a turtleneck and grow a goatie, but i'm not sure i can pull #2 off in time.
AP has a story about a teleoperated robotic vehicle to be deployed in iraq.
they already scored a 10 out of 10 in my book by using the silliest acronym since 'USA PATRIOT Act' (Uniting and Strengthening America by Providing Appropriate Tools Required to Intercept and Obstruct Terrorism). this one's called SWORDS, short for Special Weapons Observation Reconnaissance Detection System.
and yeah, 'there are no letters to write home if they meet their demise in battle.'
that's right, using actual frog muscle tissue activated by a microcontroller and a battery, swimming in a nutrient solution and driving a little swimming robot. half robot half frog.
Shortly before harvesting the muscles, two fresh liters of amphibian ringer solution were prepared according to a protocol specifically designed for frog organ culture. [...] A broad-spectrum, antibiotic/antimycotic was added out of necessity for long-term maintenance of the muscles, ex vivo. We observed, for periods greater than 24 hours, septic degradation of the muscle specimens in the absence of the antibiotic/antimycotic agents. After each muscle was placed within a Petri dish, a small volume of ringer solution was used to surround each muscle, the balance being used in the test tank for the swimming robot evaluations. The total amount of time between muscle removal from the animal to finalizing the muscle installation into the robotic swimmer was approximately 1 hour.
Muscle installation was carried out with the robotic platform partially immersed in ringer solution using #5 forceps (Fine Science Tools). After installation was complete, the muscles were allowed to acclimate for a period of approximately 5 minutes before stimulation.
After swimming the full length of the test tank, the robot was manually repositioned to the opposite end of the tank where it began, once again, to swim across the tank width. Typically, a period of swimming activity (~3 min) was followed by a period of swimming inactivity (~30 min). Due to muscle fatigue, periods of inactivity were required to restore the robot's peak swimming velocity to at least 75% of its maximum value measured during the first session of robotic swimming (first 10 minutes of the robot's lifespan).
it's a good read with few prerequisites.
for your enjoyment, here's a rough translation of what may turn out to be my final column: 'the little brothers' eyes are watching'. published a couple of weeks ago.
as always in these cases, bear in mind that this was done crudely, with no style in mind. also english quotes were translated back from the hebrew translation and might have therefore been paraphrased.
this week's column is up, entitled: 'the little brothers' eyes are watching'.
don't ask me yet where this is from, but it's too good to keep from you:
Although robots defuse or in the laboratory personally with pipettes jonglieren today long dwellings dust eyes, bombs, the way to the philanthropic art nature is far.
maybe i'm just too close, but it seems to me that robots are going to turn out to be the big tech boom of this decade.
two more robot stories:
CMU (with some MIT help) made a robot that walks on water
don't know how much of this is due to the journalist's misunderstanding, but apparently someone has filed for a patent on ethical a.i.
The patent, Inductive Inference Affective Language Analyzer Simulating AI (# 6,587,846) introduces the concept of the Ten Ethical Laws of Robotics. According to a statement from inventor John LaMuth, the patent represents the first AI system incorporating ethical/motivational terms, enabling a computer to reason and speak ethically, serving in roles specifying sound human judgment.
some robots just romp and play, others romp and play waving knives bigger than themselves.
just on time like swiss clockwork, this week's column is up, entitled 'prepare the patient, the robot is here'.
from a robots paper i'm reading:
On the other hand, we introduced a mental space with three independent parameters, the Learning System, the Mood Vector, the Second Order Equations of Emotion, the Robot Personality and the Need Model as the mental model for humanoid robots .
'sorry, honey, but i really can't deal with this right now. my second order equations of emotion are already totally wrecked today, and you're not doing enough to solve my need model'.
my old parsons mfadt classmate fang-yu lin made a robotic typewriter that "channels the invisible and intangible entity called the internet".
it's wicked cool.
like: you type something in, and the ghost of the internet rattles a reply back.
if you hate framesets like me, here's the direct link to the frame.
the first wall ever built entirely by a machine, with no use of human hands. your house is next.
the new york times runs a book review on sidney perkowitz's 'digital people', entitled 'the humanoid condition'.
in good nyt tradition it's a self-indulgent, falsely smart-looking and annoying piece of writing, with a grand finale of snobbish words in the last paragraph. but i liked this (also false) statement:
[Roboticists'] limited awareness is their downfall and their strength. The biologist makes no distinction between human and nonhuman life-forms. The roboticist takes this a step further, refusing to distinguish between living and nonliving objects.
plus, my advisor is quoted at the end...
look at those cute qrios dance (link to wmv file). sony's qrio is definitely the rage in robot animation these days.
you know, japanese robotics always makes me doubt humanity's sense of purpose and adore it at the same time.
you know, what i really like about that whole walla syndication thing is that they have talkback on their site. this way i can connect to what my readers think about my writing. in this case, my column was about a robot receptionist at cmu, and the only comment i got so far was "so does she swallow or spit it out?". ah, mediterranean class at its best.
jeff sent me a link to a page displaying a funky 22dof animated humanoid with some soccer and doggie-play-dead capabilities. as far as i understand this is currently only a platform to display complex behavior animations, but the versatility of its mechanical design is pretty impressive.
but then again, maybe this whole thing is a prank. it's getting harder and harder to tell. remember the insane transformer robot that could stop a car in mid-drive? impossible, but looking very plausible.
3d and digital video is getting so insanely good that video has ceased to be valid proof for stuff anymore. you really need to know the stuff you're looking at to be able to distinguish fact from facked.
on that note: if you haven't - you must see orson wells's f for fake ('vérités et mensonges' for ye snobs). i mean must.
well, well. who woulda thought, but this summer we will see yet another identical science fiction movie with all the same identical scenes and puns and computer graphics. and there's even a quaint mama-kid relationship with sexy hardcore action hero will smith.
just in case you want to see the by now standard 7-minute all-giveaway trailer, in which you will see absolutely every last interesting shot from the movie, be my guest.
so, earlier tonight our robot, leo, broke a link pin between his neck motor and his actual neck shaft. so we called in our very own dan stiehl who knows how to fix that sort of stuff, and he looked at leo's motors and joints while asking us what leo was doing exactly when it happend.
but less than 4 and a half hours laters, at 10:57 pacific time, the ride died prematurely in a low sputter of choking motors. the million in cash remained in the pentagon and no participating rover got to visit sin city, they weren't even close. instead a row of steel corpses lined the route from the starting point and some 7.5 miles into the desert. at the far end of which laid the red "humvee" built at carnegie mellon, an institute once named as "having the highest robot-to-human ratio in the states".
pak's short films use robots as a vehicle to unveil our inability to cope with what we crave. this can be a child, a slave to do our dirty work, or the promise of immortality. the director envisions a future in which every additional dream-come-true opens the door to more questions as to the dark side of our human nature. i wouldn't call pak apocalyptic, but he definitely has his doubts as to the future of our relationship with the tools that we create.
use filters on the right to see additional content by month or category
content here by guy hoffman .. as seen times since march 2004