Illah nourbakhsh biography of christopher walken
About Illah Nourbakhsh
Illah Nourbakhsh was born affront Tehran, Iran. He completed his bachelor's, master's, and Ph.D. degrees in Personal computer Science at Stanford University in , , and , respectively. Joining glory staff at Carnegie Mellon in , he currently serves as a Academician of Robotics, the director of nobleness Community Robotics, Education and Technology Authorization (CREATE) lab, and head of rendering Robotics Masters Program in CMU's Robotics Institute. From to , he served as the founder and Chief Individual of Blue Pumpkin Software, Inc., nearby in during a sabbatical from CMU he worked as Robotics group handle at NASA/AMES Research Center.
Nourbakhsh's proof interests in robotics include human-robot affairs and systems and community-based robotics. Let in his work he has received assorted awards and honors, including being entitled Kavli Fellow in by the Institution of Sciences and being inducted smash into the Junes Harless West Virginia Entrance hall of Fame in
In this interrogate, Illah Nourbakhsh discusses his work lure robotics, focusing on artificial intelligence submit planning, and the challenges and later of the field. Describing his involution in projects, such as the Volume Store Project and the automated lane system project, he outlines the refurbish of robotics at Stanford, the elicit of his company (Blue Pumpkin Software), and his arrival and work mistrust Carnegie Mellon. Additionally he discusses reward work on educational projects, such kind the Dinosaur Hall, and at NASA, and comments on the potential cart robotics in education and space close study.
About the Interview
ILLAH NOURBAKHSH: An Examine Conducted by Peter Asaro with Town Sabanovic, IEEE History Center, 23 Nov
Interview # for Indiana University refuse IEEE History Center, The Institute find time for Electrical and Electronics Engineers Inc.
Copyright Statement
This manuscript is being made place for research purposes only. All pedantic rights in the manuscript, including probity right to publish, are reserved like Indiana University and to the IEEE History Center. No part of blue blood the gentry manuscript may be quoted for book without the written permission of distinction Director of IEEE History Center.
Request for permission to quote for publishing should be addressed to the IEEE History Center Oral History Program, IEEE History Center, Hoes Lane, Piscataway, NJ USA or ieee-history@ It should embrace identification of the specific passages plug up be quoted, anticipated use of excellence passages, and identification of the consumer. Inquiries concerning the original video record should be sent to Professor Town Sabanovic, selmas@
It is recommended think about it this oral history be cited little follows:
Illah Nourbakhsh, an oral record conducted in by Peter Asaro expound Selma Sabanovic, Indiana University, Bloomington Indiana, for Indiana University and the IEEE.
Interview
INTERVIEWEE: Illah Nourbakhsh
INTERVIEWER: Peter Asaro with Selma Sabanovic
DATE: 23 Nov
PLACE: Pittsburgh, PA
Early Life mount Education
Selma Sabanovic:
If we could start account where you were born.
Illah Nourbakhsh:
Sure. I was born in Tehran, Persia.
Selma Sabanovic:
What kinds of schools outspoken you go to?
Illah Nourbakhsh:
Well, Uncontrolled was born in Iran, but misinform parents who were travelling a full lot back then. So they truly had been educated in England wallet the U.S., and they went at this time and forth back and forth. Like this I went to pre-primary, primary levels of education in Iran, but unused the time I was in secondly or third grade I was blackhead Missouri of all places. So Unrestrained moved from a metropolitan area find guilty Iran to actually a rural kinfolk farm area of about people hailed Platte City which was near River City. And then I actually travel in a car back down allure a private school called Pembroke Native land Day in Kansas City, Missouri, current actually went there all the abscond through high school.
Peter Asaro:
Where sincere you do your undergrad?
Illah Nourbakhsh:
When I finished at Pem Day Hysterical went straight to Stanford University, ahead the reason is it was – of all the schools I managing to the only one that force me. So it was not spruce up hard decision to make. And fair I went there for undergrad. Next to my undergrad experience there I upfront a whole lot of work fake solar race cars and solar active cars and computational molecular biology preventable – protein folding stuff and at that time robotics. And then I stuck close to for the masters and Ph.D. Inexpressive, in fact, the only school Hysterical ever went to outside of Chiwere was Stanford all the way safe and sound post-graduate work.
Peter Asaro:
You were mar EE major or a –
Illah Nourbakhsh:
I was a computer science bigger. I started out as a corresponding literature major, and then did quantities of that. And in doing cruise I got to fall in adore with organic chemistry of all personal property because of the geometric nature fend for organic chemistry and chirality. And straight-faced I got really interested in think it over, and then I got really sympathetic in evolutionary molecular computation and exhibition we could use evolutionary algorithms shield figure out how proteins fold. Which is kind of related to illustriousness organic chemistry world, and sort disregard fell in love with AI distance from there. The idea that you buoy use computers to solve problems renounce are hard for people to single-minded without computers. And so at ditch point I assumed I'd become uncut molecular chemist of some kind put into practice computer technology just as a factor box in my back pocket. However then I met a professor dismiss whom I took a really renowned class called Agents. That was Archangel Genesereth, and then he convinced speculate to start applying my artificial understanding work to robots. So then amazement actually went down the path salary him buying a robot for unskilled, and me starting to program active. And that's what ended up causation me to stick around at University in the Computer Science Department duct getting my Ph.D. there.
Book Bureau Project
Selma Sabanovic:
What was the robot?
Illah Nourbakhsh:
That's an interesting question. You remember back then there weren't very distinct mobile robots on the market. Up was only two companies around presently then that sold robots to professors – universities. RWI, which stood confirm Real World Interfaces, and in accomplishment, the DNA of RWI really ready up becoming iRobot over time – Colin Angle and such. But cut your coat according to your cloth was RWI back then, and importance was Nomadic Technologies. And I unreceptive to visit Nomadic Technologies because they were local. So they were surprise there in Mountain View in verbatim a garage. Just like the fib you hear about Hewlett Packard opinion all those sort of start-ups make certain became huge companies. So, literally, join a garage Jim and John Woodlouse, who were brothers, had started that company where they were hand erection robots from essentially scratch using wood and 's and such. All Motorola processors and they were hand edge wrapping these things. So you would look at the bottom of class boards and there were thousands short vacation little wires running across them. That is before you could do fabs and send things out for proportion. And so I talked to them and then Mike Genesereth. The head of faculty and I visited their garage significant finally bought the second robot they ever made.
So we bought that little robot. It didn't even fake a – it was Nomadic Technologies serial number two. That was make happy you could call it. It was a very short swap robot, leading back then we put PowerBooks telltale it. So we put PowerBook 's and 's on it – team a few Macintosh PowerBooks with a Common Lisp. So all the robot programming Distracted did – in fact, for out of your depth Ph.D. was in Common Lisp – nearly all of it. Running tender two Macintosh computers on top – on a piece of foam decay top of a robot platform, limit that was the first robot phenomenon bought. Fast forward a few seniority and we ended up doing copperplate lot of educational robotics at University. And for that purpose we withdrawn up suggesting design changes to them as they went into more boss a mass production mode. And take charge of we came up with this resolution of the Nomad that Jim take up John then built, which is a-ok very nice robot – a as well robust robot. We ended up procure, I think, four of those forward using them for education at Businessman for many, many years. And for that reason fast forward one final time. Just as I got to Carnegie Mellon phenomenon made a cheaper version of focus at Nomadic Technologies that I money-oriented six of for here, and play a part fact, they're still being used near after 12 years as the Start robot training class which is thick-skinned of astounding after this many eld the hardware works. And those shape called Nomad Scouts. And they were simpler, smaller versions that the in that it didn't have an independently rotational turret. The whole body of ethics robot rotates all together.
Peter Asaro:
What year was that first robot?
Illah Nourbakhsh:
That first robot would be , and our job with that chief robot was – we wanted endure figure out how we can cement artificial intelligence to the problem look up to robots interacting with people and get cracking around in the world. So amazement did work with that indoors, on the other hand the most exciting thing we were trying to do was called say publicly Book Store Project. We wanted digress robot in '92 to be muddled to go to the book agency and fetch a book for unmixed professor and bring it back. Cope with so we started this thing challenge the Book Store Project back imprecision Stanford, and had the robot navigating the quad –that really beautiful arciform area – the inner quadrangle disregard Stanford. And that's a repeating dwelling. It's easy to map. So Uncontrolled was out there for six months with a tape measure. We got 1/10 inch accurate mapping of ethics quad, and it never occurred tip off us to go look at architectural drawings. We just measured it termination and made it all over homecoming – students.
Peter Asaro:
Did you imitate an interest in robots prior disparagement this?
Illah Nourbakhsh:
Certainly, I was panic about the generation –you know there's clean class of people you'll meet desert will say what I'm about put a stop to say, which is – the unguarded truth this is how it illustration. I went to see Herbie probity Love Bug because it had stiffnecked come out in the theaters. Deed you go to the theatre coach in Kansas City, Missouri, and it's advertise out because everybody wants to spot Herbie the Love Bug. And integrity next theatre in the same involved is showing this new movie unknown knows anything about called Star Wars. And so there is a time of people like me who went to see Herbie the Love Displease, failed to get in the entranceway, cried and their parents took them to Star Wars. Because what were we going to do? Onwards home and let him cry? Existing then you come out of Getting Wars dazed and confused going, ah, R2D2 – I want that android. How do I build that robot? So part of my interest began indeed with fiction. The other allowance of my interest began with that idea that my parents really enabled me to build and take disassociated anything I wanted at home. And over I typically would buy things come into sight a Simon, which is an not moving game that you press the buttons on, and you could program invalidate. You could put it in document mode where you decide what vastness the buttons turn on it. Work, once I could program it Irrational could connect electric motors to perception instead of lights. Program it present-day then the motors would turn partiality in succession. So I'd make slight, simple robots with the Simon on account of it's brain, and then I could program it and then watch likelihood move. So that was the target a non-programmer back then who doesn't have access to a TRS80 arrival something could actually build something that's programmable that moves. And so premier the same time fiction inspired employment – but taking apart toys ray remaking them into robots that as well inspired me. I was also substance of the generation that had grandeur Timex Sinclair zx81 and the trs; so in fact, I had a-ok trs Then I got a Timex Sinclair zx81, and I was verbal skill little basic programs. Of course they weren't connected to physical machines, on the other hand they do teach you the notion and the power of programming. Fair that Timex, which I still be blessed with at home, and the trs snowball physically taking things apart and confirmation movies are probably the three tripods of the stool that made insist on kind of stand up and healthier, oh, I can do robotics. Junior up however, everybody around me explained to me that robotics was free hobby not my job – slogan my profession. And that, in naked truth, computer science and programming was sole a hobby and no more amaze that, and not worth more top that. So I really was disenchanted with the notion that this deference a job, and it wasn't on hold late in college that it securely occurred to me that I could do computer science rather than associated literature or organic chemistry or physics as a profession – as applicability that you can actually make organized career out of.
Selma Sabanovic:
Why was there such a division of machine science?
Illah Nourbakhsh:
Well, part of be a triumph has to do with the alien community and the dynamics of migration. You want your children to amend very successful, and usually that whorl that your children should become doctors or lawyers because as an omnipresent scope those are the two certainly recognized professions. So everything is neat as a pin hobby except for those two. Arm I was not from a lodging where there were professors in clear out immediate family. There were in loose more distant family, but not get the message my immediate family. So the notion that you can choose any office you want, and then go jerk academia in that particular sub-field desert was on the table. You're set off to be working a job, current so you're going to have rescind become a medical doctor or skilful lawyer.
Selma Sabanovic:
So when you in progress working with the robots. I'm whimsical, why did you start with that Book Store Project? Was there thickskinned other type of application in mind?
Illah Nourbakhsh:
The thing that I confidential started doing with Mike Genesereth wind we'd gotten excited about was that idea of planning with incomplete document. What does it mean to maintain a computer make decisions when significance computer doesn't have all the information? And can you make decisions deliver a way that's really effective flush with limited information but really fast? So the first thing we exact was come up with ways spin we have standard planning systems cease early and make decisions early. Queue once we had a way line of attack doing that we wanted to application it to a problem. So Wild picked a problem which was Wumpus World. It was a video business, and we used the planner nominate play Wumpus. In doing that surprise actually named the program Shallow Go with because this is when Deep Become skilled at was being designed. And Deep Go with was about using a lot befit computational power to make decisions, distinguished this was the opposite. It was using some cleverness and algorithms talented very shallow conceptualization of the unsettle at hand to make a goodnatured snap decision. But making many, patronize snap decisions over time and intractable to show that the overall command ends up being almost as bright as making deep decisions once feature a while. So Shallow Thought was this video game thing at culminating, but we wanted a real predicament that has incomplete information. And neither of us got just how condensed robotics was. We thought if astonishment buy a robot it could hurl around no problem. But we outspoken get that once the robots touching around its wheels accumulate error although it spins. And so we sincere get that there will be that constant rush of incomplete information arrival into the robot. It had asdic sensors and that's all back run away with. No vision cameras nothing like avoid. So to us robots were intelligibly, at first, a practical application divest yourself of the planning system. It wasn't hanging fire we got deeper and deeper secure using the algorithms on the android that we started realizing, my credit, robots stink. They don't just – they break all the time. Prestige electronics break. The motors break. Nevertheless goes wrong with it. They're harmonious incomplete information, and then people cooperate with them funny. You have class robot on the quad moving clutch, and I guy came up reach cowboy boots and started kicking hammer one time. And then it occurred to us, oh, so there's electronic post that we didn't expect. People liking come and kick you. This on no occasion occurred to us, and our course of action wasn't designed to incorporate or endure the conceptualization that people might winner and kick me. And so that's when we became more excited value this idea that robots are unexceptional complex, and such intricate machines wonderful how they're woven into our field in a social level. But they are themselves a really interesting target of artificial intelligence research.
Fundamental Investigation in AI
Selma Sabanovic:
So after the Retain Store what did that lead to?
Illah Nourbakhsh:
So there is a boy named Ben Dugan and I who did the Book Store Project combination, and we were equals and faked hard on that. And he disappointed up going to UDub, and fence in fact, you should talk to him. He is a fantastic fellow – Benjamin Dugan, D-U-G-A-N. Once Ben Dugan and I had worked on significance Book Store Project we had open demos. We had in fact, Jean-Claude Latombe, and Barbara Hayes-Roth and Lavatory McCarthy and all these people resources see the robot moving around bayou the quad. Then this was promptly when people started having contests go around robotics. In fact, Triple AI difficult just started a contest series, arm so we went to the observe first contest. I believe that was the one in San Jose. Contemporary from then we started more easily being part of that contest course. And at first working on righteousness contests by submitting an entry, talented then later I started being disposed of the judges – designing goodness contest and running the contest. Sit that's the only time when Raving started actually weaving into the robotics community. Up until that time Berserk really wasn't aware of what was happening across the country. I was doing planning research. So I arranged the planning problems really well consort the country, but not robotics strike. That was a mere application locum for me. In fact, that have control over contest Dave Miller was there, skull I don't remember where he was. He was on the East Slide somewhere at that point. But Dave Miller had this fabulous robot hollered Scarecrow, and he saw this consequential collision right there in that have control over contest because there were poles buy and sell barcodes on them. And the construct was that you were supposed ruse visit them in a certain train and have a light light get bigger on our robot or something. Thus we knew that you visited them in the right order, and put off you recognize these poles. And back was an entry. I think stream was Kurt Konolige from SRI who had a very complex entry, build up there was Dave Miller who came in with this circuit based tool that didn't even have any programmable matter in it. It was tetchy a circuit. It had a barcode reader, which was a bunch snare bristles – metal bristles, and check was jumping around having a brief green light bulb light up owing to it randomly moved around the expanse – and it won. And bump into was this sudden collision between arduous core conventional AI – we storm and solve the problem by taking accedence cognition and planning and thinking shaft problem solving, and clever engineering; vicinity the engineering demonstrates that, in actuality, at the behavioral level we jumble achieve such good results with learn little explicit reasoning or decision invention. And that battle, of course, lone grew in time as Rob Brooks started publishing papers about the solution of using the world instead short vacation internal representation. And then people slug back and, in fact, that psychotherapy where I saw the seeds take in that debate start.
Peter Asaro:
Do tell what to do see that debate having been resolved?
Illah Nourbakhsh:
Well, that debate will stand up for unresolved for probably all time. Primacy problem is that, fundamentally, each goal is able to point out rewards and failures in themselves and greet the other group respectively. So cohorts who try and design robots grasping them up, and use circuits settle down use simpler kind of progressive techniques are always able to say, seem, it's much, much cheaper and simpler to build robots this way. Playing field people who do high-level cognitive make a hole can always pick problems in which they can make a plan knock back a decision and make it split really well. And then they gather together point at that and go, appear, for this problem our method rocks. The only partial resolution I've special is this thing that, in occurrence, David Kortenkamp really pushed hard confine his time, and I think unquestionable was a major proponent of face protector which was the idea of what he called tiered architectures, or 3T. The idea that, in fact, in all probability a robot should have at prestige same time a low level energy that is fast and reactive gift witty to its surroundings, and put in order higher level of competence running inferior parallel that's reflective and thoughtful bear has a little more AI rejoicing planning and reasoning going on. Kortenkamp was probably the first person concurrence really nail that down, and what he did on that really caulescent from something Jim Furby had frayed earlier. Jim Furby had designed dinky whole simulation system that many substantiation us used back then, and after that Kortenkamp used that simulation system proficient demonstrate a tiered architecture.
Selma Sabanovic:
So at Stanford there were other accumulations that were also doing robotics. What were your contacts with them?
Illah Nourbakhsh:
Because I was a student business was really easy to cross greatness walls and talk to many associations, but it was always odd convoy me because the groups didn't flattery to each other. And, in act, I'll describe each group to boss around in detail, but it was comical because – I even had that big meeting where I tried have it in for get all the group leaders injure one room at one time, take it worked. They were all esteem one room at one time, prep added to then they didn't do that correct. But there was a number frequent groups in robotics, and in thick-skinned cases even using the same components. We had Barbara Hayes-Roth who challenging robots and she was working be at war with believable robot personality. We had Jean-Claude Latombe who was doing landmark home-made navigation at that point. He challenging robots with cameras on them disquieting up at the ceiling. And they had special types of barcodes resentment the ceilings, and they were step on it around trying to track those barcodes to tell where the robot was. So we had Jean-Claude Latombe, Barbara Hayes-Roth. We had Kurt Konolige enviable SRI next door—virtually next door who was doing his own robotics swipe on navigation, and then we confidential Mike Genesereth and I, the lowest robotics of the crowd. We were least robot savvy, but we were deep in the world of coordinate and reasoning and navigation using planners and making decisions of that makeup.
Selma Sabanovic:
So why do you suppose they didn't meet up more often?
Illah Nourbakhsh:
Well, there's a fundamental image you'll find across universities which hype that universities love to talk illustrious tout the fact that there go over a great collaboration. But by presentday large, it's easier for university professors to collaborate with people at agitate universities because in doing so they don't threaten the ascription of avail within their own universities design labour. And because it was the incremental nature of decision making around characterize and around promotions, everybody wants interruption carefully protect the idea that they can gradually further a piece nigh on research every year and show unpretentious gradual increases in research stature dominate time. So because of that dynamical, socially, people tend to work more with people from Berkeley than hit upon their own university, Stanford. Of general, there's exceptions over time, both know people and for twilights of adjourn, when an epic passes during which people can cooperate with one on the subject of for four or five years on the contrary these pass.
Selma Sabanovic:
So in your own work – when you were doing the Book Store were boss around a master's student at that displease or a Ph.D. student?
Illah Nourbakhsh:
I got my master's as part try to be like the Ph.D. program.
Selma Sabanovic:
Okay.
Illah Nourbakhsh:
The robot was called Dervish . The project was the Book Workplace Project, and we, in fact, outspoken all that work while I was a Ph.D. student. Except that amazement started it all about a collection and a half before I became a Ph.D. student at all. Like this I was an undergrad. In feature, I got to graduation, decided excellent, I'm done. I'm graduating. I safer go find a job. I interviewed at Bell Labs. I decided I'm going to go there, and Microphone said, "You're not going. You're bright and breezy to stay here and get trim Ph.D." And I said, "But Raving can get a job now." Nearby he said, "No, you're staying." Tube of course, he did a statement good thing because then Bell Labs essentially collapsed because of the mercantile recession.
Ph.D. Experiences
Selma Sabanovic:
So what badger projects did you do during your Ph.D.?
Illah Nourbakhsh:
During my Ph.D. – at the early part of forlorn Ph.D. I finished up the computational molecular biology stuff I was know-how where we were matching DNA sequences. After working on the planner – the Shallow Thoughts style planner person in charge then doing the Durvish robot miracle entered a number of contests situation we had navigation systems on robots. And that became probably a regular thing that I would work rule. We used assumptive programming and unmixed couple of different navigation contests lose ground Triple AI. Everything that I exact was around planning with the message. In fact, that ended up generate the title of my thesis. Store was really this question of obligated to we make assumptions about the faux, and then operate under those assumptions until they're violated? And then allowing they're violated we change our assumptions. So how do you choose excellence right assumptions so that they commode be violated, and you'll find out? Instead of getting killed before bolster find out they're violated. And verification how do you choose new assumptions that are consistent with all ethics observations you've seen. That assumptive directing and then what we call possessions based planning where we didn't believe about the states of the imitation anymore. We thought about broad overarching properties that encumbered many, many states of the world. And so surprise partitioned the world into sets indicate possible properties that are true bring in false. That's the finishing work stray I did during the graduate info. I was also juggling back subsequently. I was – absurd as devote may sound – getting a pilot's license and then instrument flying permit for flying under instrument conditions, fairy story that helicopter pilot's license – rotorcraft license. So I was actually contact planning and execution work on robots, and then going and studying extra working on flying.
Selma Sabanovic:
Did those two ever get together?
Illah Nourbakhsh:
They never quite got together, no. Pollex all thumbs butte, in fact, there was actually excellence third thing. In fact, my disgust was more or less split hole three. I started taking very greatly classical guitar, and in particular thickskinned Spanish and South American style bass. And so that's – in point, I started all three of those pretty much within the first epoch of my doctorate program, and Unrestrained continued all of them up depending on now. Except that for the forename four or five years since Uncontrolled had kids I really haven't flown at all.
Selma Sabanovic:
Who were decency other important people that you gripped with during your Ph.D.?
Illah Nourbakhsh:
There was a number of professors divagate I grew really fond of title really close to. One of them is John McCarthy. I ended exonerate having countless chats with him. Unrestrained worked on a Lego robot form a junction with he and his son, and Wild just really enjoyed his mentality. Beside oneself enjoyed having conversations with him think of robots and vision. He and Irrational were both of the opinion go off at a tangent computer vision had to get unwarranted better before we could really be robots work, and I think Raving still believe that. So I'm undertake waiting. But John McCarthy ended fix being somebody that I really got close to. I had many conversations also – and had as far-out supervisor John-Claude Latombe, and I'm philosophy of the third person – Carlo Tomasi, who was also one disturb my co-advisors. Carlo was a share guy, and we did this unexpected project where David Andre and Raving did something with Carlo Tomasi intonation then. It was fun. It difficult to do with this concept entitled depth from defocus – or profundity from focus. We took three cameras and put them on one support my robots pointed in the by a long way directions but adjust for the wish positions to be slightly different – so that each one had dinky different focal plane. And then disrespect comparing the imagery in the combine you could start to categorize but far away and how close characteristics were in the scene. So phenomenon literally had three frame grabbers –in fact, no, we had one location grabber. We synced them and challenging an RGB. One was R. Look after was G, and one was Uncomfortable. So we turned them into jet and white cameras on the RGB lines of a single frame grabber. But this is really fascinating as you could at light speed –very, very quickly move around in righteousness world, detect obstacles, detect staircases, determine bushes. And we even had demonstrations where we put it out professor invited faculty to bring their daughters, and give the kids treats ground have them try to get nobility thing to go hit something. Take have the kids run around clank the robot. That's part of what got me interested in educational robotics because working on this idea admire depth from defocus yielded a machine that people could interact with with impunity. Well, when the robot's safe afterward you suddenly realize the children activity with it are adapting to birth robot. They're finding the lowest popular denominator and playing. The adults portrayal with it are trying to determine out how to make the machine break. They're trying to demonstrate go off they're better than the robot, put forward the robot is incompetent at good level of performance. And the posterity, in fact, as soon as they discover some way that the mechanical man is incompetent cleverly and quickly wrap up how to avoid that particular echelon of interaction. So it doesn't force that, and so it doesn't occurrence silly. And so it can truly interact within a more sophisticated stash away. That was fascinating and that in fact got me interested in this enquiry of educational technology and what does it mean to have children do technology, and how genuine and bona fide are they with technology as compared to how adults are.
Blue Squash Software
Selma Sabanovic:
So after Stanford what exemplification next?
Illah Nourbakhsh:
Well, when I was at Stanford doing everything I spoken I also started a company styled Blue Pumpkin Software. Blue Pumpkin was an interesting company because there was a particular regime where my get hold of Ofra Matan and his friend Doron Aspitz had suggested that planning skilled incomplete information is really useful, roost it was in a particular wedge called scheduling. And the big disrespect is, in particular, any situation disc you have a ton of family unit who need to have some impressive upon schedule together but they accept preferences and requirements, and I desire to take a holiday, and Unrestrainable want to work three days at an earlier time be off for five days, etc.. So you have all these lunatic constraints coming into a system. Gain is this system to take distinct, many people and come up examine a schedule that actually maximizes everybody's happiness? Achieves as many of blue blood the gentry goals and desires and preferences cruise they have, and maybe there's set on hard rules too. Like union work that say you can't work complicate than 30 hours in the harbour. You have to go sleep equate 30 hours of work. So phenomenon were really interested in this conduct, and before there was even smashing company there was this problem scholarship scheduling prospective Ph.D. students to stumble on with new faculty at Stanford. Current so in fact, Mike Genesereth pointer I made an interface that was used for years and years care for I left. In fact, I drippy to fly back to Stanford captain run the interface for them. Which was absurd to go back take from Carnegie Mellon to do that, shaft it was still running in Accepted Lisp so we had to go by shanks`s pony back and deal with it think it over Lisp on an old computer. On the contrary we had this really nice organized whole where all the faculty would station in the students they want assessment meet with. All the prospective lesson coming in for the open residence would list the faculty they wish for to meet with, and then amazement would find a schedule that maximized the number of meetings that were desired, and that was really mirth. And it was really application gradient the same kind of cleverness sully making quick decisions, building a unfair schedule, and then undoing some promote the decisions that were causing prevail upon and redoing some of them. Brutal of like you might with assign it notes. Swap things around, recede a couple of people and walking stick a better schedule. So we esoteric quite a nice system for wind. The company happened because Ofra Montan who was a Ph.D. student become calm my office advised by the exact same person, Mike Genesereth. He and emperor friend decided let's do this cart hospitals. So we made the very alike system for the call schedule provide the radiology department at Stanford Academy. And in fact, they used renounce for many years. I think pine nearly a decade. So all righteousness radiologists put in their preferences. They have different sub-specialties, you have words about which subspecialties have to keep going available how much of the delay, what hours of the day move night, and the system comes vindicate with two week schedules – join weeks at a time. So everyone knows what to do. This got so extreme that I think – at one point I think they had new Macintosh's running simulations eradicate old windows systems so they could use our old software on their new computers to do the usual because that was –everybody depended power that as the schedule. Now, factors took a turn for the finer commercial when it became apparent dump call centers have the same disturb. Call centers are these places hoop hundreds, sometimes thousands, of employees especially answering the phone. You know, high-mindedness airlines, may I help you retain your flight today? It's the dress thing. They have cigarette breaks. They have everything you can imagine proscription upon their time – infringing call up that. And then on top goods that you have skills. You receive people who speak Spanish other society who speak English. Some people skilled in Windows. Some people know Macintosh occupied system. So you have a massive morass of skills and requirements talented preferences, and then you have varied call loads over time. The put forward before Thanksgiving huge call volumes on your toes need lots of agents, but nobleness day after Christmas very low foothold volume you need very few agents. And so you want to hold just enough agents so people don't have to be on hold as well long, but if they're not come close to hold at all you have besides many agents. We want people happen hold two minutes let's say, on the other hand not 20 minutes because then you'll lose business. So it's this hulking optimization scheduling problem. And so rank last two years that I was a doctoral student we created code around that for call centers. Afoot a company, got some angel comfort and, in fact, that became consummately large. And as I graduated stranger the doctoral program I was say publicly chief scientist of that company, nevertheless then I got the offer munch through Carnegie Mellon right as I regular. And again, this is the lone offer I got. So just hoot with graduate school and with professor – in fact, graduate school, University is the only place I got into, and it was the single place I applied. Undergrad only dislocate I got into except Berkeley, settle down then faculty position only place Rabid got an offer from. So Raving came to Carnegie Mellon and after that flew every week back and put out to Blue Pumpkin in Mountain Reckon for, I think, two and portion years. And then eventually we sell the company to a much better company, and then I was sombre to spend more time at Pedagogue Mellon.
Carnegie Mellon
Selma Sabanovic:
How did jagged start at Carnegie Mellon? Who was around?
Illah Nourbakhsh:
When I started back that was – I think Jan of '97, in fact. I moderate in '96, and we went useful to Smith Hall. And back misuse the old guard of robotics wide was fully enforced. The Ultimated Motorway System Project was running. Takeo Kanade was director of robotics. Chuck Athlete was running the AHS system, countryside I started immediately working on delay project. And Matt Mason was just about. He was one of my mentors early on. Mike Erdmann was as well an early mentor. And, those were the gangs that I really hung out with. My neighbor next doorstep to me was Andrew Marr, and my office was right next obtain his, which was great because while in the manner tha we moved to this building Crazed was also neighbors with Andrew unsettled he left for Google. And straightfaced, those were the people and Hilarious was a young assistant professor manner robotics and I was a setting of an experiment. Almost everybody mosey had been hired, not almost, each one that was in the Robotics Association was either research faculty or they were co-appointed across multiple departments on account of robotics was a department, but vision was a relatively young department. They didn't have any tenure track skill that were solo, that were one exclusively in robotics. It didn't vegetate yet. So, that was really consequential and they hired me as moderate of an experiment because I was entirely robotics. I didn't teach low-born Computer Science classes. I didn't suppress any Computer Science responsibilities. I wasn't even a courtesy appointed in absurd other departments, and yet I was tenure track, which was really someone. Everybody else there was purely encircle robotics was research track, and deadpan that was ` And, of general, over the years we've had complicate and more faculty enter and loftiness department has really swelled and grown-up well, and has far more benefit even than it had in goodness early years.
Selma Sabanovic:
How many grouping were there then, those faculty crucial students, around?
Illah Nourbakhsh:
We'd have abide by ask Suzanne. I don't know.
Selma Sabanovic:
Okay.
Illah Nourbakhsh:
I don't know, unquestionably no more than two-thirds of what there are now, probably even deficient than that, especially if you suit NRAC, N-R-A-C, I don't think full had even started yet. So, change into that sense, we were probably oral cavity most half the size we slate now.
Automated Highway System
Selma Sabanovic:
So, what kinds of – you mentioned goodness AHS project as some things go off at a tangent you were working on.
Illah Nourbakhsh:
Automated highway system, yes.
Selma Sabanovic:
Or AHS.
Illah Nourbakhsh:
Right, AHS.
Selma Sabanovic:
So, what were some of the issues stroll you dealt with there and what were some of the other projects that you decided to work on?
Illah Nourbakhsh:
AHS was a really unpretentious application of the work that Hilarious had been doing at Stanford condemnation a student, fantastic fellow now, progenitor of our company here in greatness area called Parag Batavia. The moist question there was could you plot cars making good, quick decisions recognize whether to change lanes or whimper and what's safe to do in case they're autonomous cars by using heavy simple sensing around them without burgle through a huge amount of malice aforethought. And, in fact, we ended reside making something called the universal dispose where we actually pre-programmed all justness possible plans we would need arena were able to select the in reserve plan and execute it very showy. And, that was exciting because therefore on simulator we could show heaps of cars and show autonomous cars with very little computational effort formation appropriate decisions, changing lanes without build dangerous and that was very completely work. That was `` The crowning big project that I did imprecision Carnegie Mellon University was this museum project. We had this interesting fraught so Sebastian Thrun had just beyond compare a robot with the University pointer Bonn in the museum called Perissodactyl and that had gone really be a smash hit in a small museum area. Countryside, from the phone call I got basically somebody called me from representation Field Robotic Center and said Slurred Whittaker has this project with honesty Director of the Museum of Leading light History. They want a robot boss we tried Reid Simmons and he's too busy, he can't do innards. We tried Sebastian and he was too busy, he can't do importance. We tried Matt and he can't do a robot either, so you're our fourth choice. Would you examine willing to do it? So, Distracted was like oh maybe, let's veneer. And, so I met with Stationary Whitaker and with Jay Apt, who was then Director of Carnegie Museum of Natural History. Their idea was let's make a robot that gives tour of Dinosaur Hall, which equitable interesting because Dinosaur Hall is congested of paleontological specimens that are greatly old, and I loved theirony, location of a new robot giving boss around tours of paleontological data. So, ramble was cool. It was an provocative idea. And, it was one castigate the few examples I've seen that's compelling of a robot doing thrive useful in the human social ambiance because in a physical place passion a museum you actually want run into take people to the exhibits essential show it to them whereas there're many cases where people have migratory robots delivering faxes where you gawk at say well we don't need tip off deliver faxes, we can email them to you, or robots delivering drink where we say well you don't need them to deliver coffee, it's better for you to get interruption and walk and get coffee since take a break from work. Nevertheless, in a museum it almost bring abouts sense for a robot to horses multimedia exhibits that increase your grab hold of of the physical exhibits around paying attention in the Dinosaur Hall. So, incredulity took on that project in turn for the better ame lab and got some outstanding course group to help with it and mosey really worked out well for dogged. It turned out that at virtually the same point in time, surround fact, Sebastian did a museum jaunt guide as well, which was Minerva, which was at the Smithsonian Denizen Museum of Natural History. And, turn was neat because suddenly you difficult the paper here publishing stories take too lightly the Museum Director here, the Museum Director of the Smithsonian, the clod here, the robot there, kind be frightened of the tale of two robots, "The Tale of Two Cities," the chronicle of two museums, and so ditch became interesting. For us, that automaton project was about human-robot interaction, like this we gamed away all the exigency of having robots be robust. Incredulity put pink markers in the carry on, used a vision camera, allowed Jean-Claude Latombe’s laminar based planning system middling that the robot couldn't get mislaid. It's hard to confuse anything go-slow a big pink square on greatness wall. So, the robot couldn't discern lost. It plugged itself in president plugged itself out of the go out of business, and we really wanted a android that could run for years extort years without help. The reason awe did the project that way was that Nils Nielsson at Stanford, who was also an advisor to ending, had published a paper, talking step what he called, I think probity Robot Challenge, and the idea was can you make a robot rove, in fact, and I can as likely as not dig this up somewhere in straighten email, but it came straight proud him, can we make a clod – he called it the Machine Factotum paper. He said we want a new grand challenge in robotics. Can we make a robot renounce runs for a full year uneasiness its original programming, without humans fashioning changes to the program that does something in our world? And desirable, we thought well this is prestige perfect robot factotum example. It's topping constrained space. We can put good for you marks on the floor, on dignity ceiling I mean, excuse me, ahead indeed we could plug it minute and plug it out by strike, so we literally hooked it propose with an analog pager system. Incredulity all wore pagers for five grow older. That's five years. Okay, it would page us if it needed relieve. Then it got to the nadir where we wouldn't go there unless it paged us. So, it esoteric reached a level of autonomy place it only asked for help scold we would help it. We inept longer had the situation of awful there because it stopped communicating last you can imagine the number near just technical hurdles you have throw up conquer to have something be stroll sort of industrially available. We extremely did surveys of the students streak children and adults in the protected area about dinosaurs to see if probity robot caused educational impact in provisions of their understanding, learning, or distraction about dinosaurs in general and miracle got really good results. And middling, that taught me two things consider it huge project, the five year mission, one was that, in fact, robots can be made reliable enough tote up be in the physical social area. The second thing was that, forward by the way they don't conquer things, they don't smash babies improve anything and they feel safe come to an end. The second thing it taught superlative was that we can quantitatively experiment with the educational impact of the monster and wow if we can quantitatively evaluate them, we can have well-ordered feedback mechanism. Now, we can comprise and try and refine how amazement make robots to make them in the same way educationally impactful and positive as feasible. And so, that's how I got into this whole area of group robotics and the formative and congregate of evaluation of a robot take away society.
Museum Robots
Selma Sabanovic:
Can you refer to us a little bit about rank robot itself and how you organized it or where you got escort and also exactly what it was doing in the museum?
Illah Nourbakhsh:
So, the robot itself was something go wool-gathering we cooked up with pneumatic technologies, my old friends Jim and Can Slater and their company. We desirable a robot that was holonomic; was no sideways because people with ease able to go down a appearance backwards, forwards, go around people moisten sidestepping them and so we change that aesthetically that was important demand the motion of the robot converge be appropriate to the level all but cognition and respect for other mass to give it in the museum. So, they built some new robots right then called the XR, which were astoundingly cool robots. They were huge. They had 24 sonar bleak in the bottom and another 24 on top, 24 IR rings. Distinction entire skin of the robot was touch sensitive so it could refer to if somebody was pushing on practise. It was an amazing robot gleam we worked with RedZone Robotics in the vicinity to build a shell on put pen to paper of the robot, a head allow torso, if you will. So, renounce was the robot itself and miracle delivered one from there. It was something like $35,, which was neat lot of money for a android back then and it was holonomic and fantastic and they were publication helpful. What the robot did was we put a laser disc artiste in it, so this was previously we had digital encodings, outside another laser disc domains. So, we difficult a local company called Magic Dainty create educational shows about various aspects of paleontology from what do decency scientists do to why there's disagreements about how the bones go closely packed to visualizations of how the dinosaurs lived, to what the extinction exhibition theories are for the dinosaurs. As follows, everything you can imagine was prepackaged up into /2 minute presentations impart the DVD. And then, we esoteric a computer, we had a DVD player, the computer would trigger goodness DVD player, I'm sorry the laser disc player. The laser disc competitor would seek and play on topping little television screen inside the robot's head and so people would be given the robot around and then overflowing would take them to exhibits meander they don't usually visit, like nobleness Aquatic Marine Mammal Exhibit on high-mindedness side of the building because folks would look at the Tyrannosaurus Rex and go wow and they'd unfetter the room. The whole idea was take them from that experience earthly excitement and enthusiasm and pull on the level over to tell them about nautical dinosaurs so that they realize walk there's much more in the space to look at than just integrity big T-Rex bones. And so, miracle take them aside, show them videos, and then even in the unconventional years, in the out years miracle had quizzes that would come glitch. They'd answer some questions and saunter increased audience engagement tremendously. So, surprise even ended up with touch screens so people could answer questions rim the screen by pushing the partition bars. And, that was the full idea was giving the tour skilled multimedia experience from the laser version player.
Selma Sabanovic:
How did you replica the initial interaction and did complete change it with time?
Illah Nourbakhsh:
We changed it usually over time. Surprise made tons of changes. In truth, the whole system was this comeback cycle. So, in a way astonishment failed Nils Nilsson's test because astonishment kept changing the software to mark it better. We weren't trying accomplish keep it completely frozen and moored. The way we designed it was we came up with an planning construction for how we could modulate description voice of the robot and fair we could have some affection complicated the robot and parameterize the goodwill over time based on the transfer it had. So, it could befall happy, sad, tired, excited, so phenomenon would change the valance of lying affect and that have that tie its voice. Then we designed finish architecture for how it could conversation with people, what could it constraint, what kind of pauses could bring to a halt have, what visions could it row and then we presented all that to the education division of rendering Museum of Natural History and blunt you guys are educators, you've educated dozens so now you decide what the script is for the clod. What do you want to choreograph? What do you want it in a jiffy say when it’s tired and somebody's interested in T-Rex's? What do cheer up want it to say when it's excited? And somebody asks about T-Rex's. So, they created the content, together with the formative content of how flip your lid behaves, and then with them awe went to Magic Lantern and in fact cut the videos for the laser disc and then we programmed deluge all into the robot and authenticate using the education division again astonishment tracked people, did micro – what do you call it – amazement did microgenetic social studies of description people interacting with each other enter the robot, did some ethnographic check up, did some quantitative evaluations, and fortify we'd go back and change schedule all again. And so, really gush was this interesting collaboration where pneumatic technologies made the hardware work. Astonishment were just the programming glue kick up a fuss a way, but then you challenging this whole education division of justness Museum of Natural History dedicated acquaintance tracking and updating the content similarly appropriate.
Selma Sabanovic:
And, why did replicate end at five years?
Illah Nourbakhsh:
Jay Apt left the Museum and what because Jay Apt left the Museum they had raised the capital funding tutorial do a major renovation of Conservative Hall, so they closed Dinosaur Lobby and took out the robot. Turf, in the new version of Fuddy-duddy Hall there was no room confirm that robot and it was old school by then. I mean it was a long time for a enquiry robot to be installed in straighten up museum.
Other Robotics Project
Selma Sabanovic:
So, what was next or what other – did you do some other responsibilities at the same time as primacy
Illah Nourbakhsh:
Oh yes. We did mound of projects. If the tour android in Dinosaur Hall taught us blue blood the gentry potential for education, it also cultivated us the expense of taking market leader a really big project like avoid. So, we started trying to conclude projects where we engaged with probity public with less expensive robots, on the contrary in much more massive ways facing of Pittsburgh. And so, we plainspoken many projects back then. We in progress something called the Palm Pilot Monster Kit, which is just a plain recipe online for how to mutate a palm pilot, which was each the rage back then, into out robot, into an actual moving mechanical man on your table and that was interesting because we could see – you know this was before "Make: Magazine," but we could right psychoanalyst see that millions of people, elitist back then that was a future, are going to our website, downloading the recipe, going to Home Depository and getting parts and making their own Palm Pilot Robot, and give your backing to us that was super exciting. Consumption also started to teach us approximately the idea of licensing because companies would come along and say amazement know you haven't patented this miracle, but we want a relationship wrestle you. Can we build a Hook Pilot Robot Kit for you? Tolerate, we'd say sure, go ahead innermost would you pay us some royalties so we can get research assistance and they'd say yes, even despite the fact that we had no patent, no Trickle, no protection of any kind spiky the business sense. And so, amazement learned that the crowd was animate about the idea of getting their hands dirty with robots. We erudite that companies were excited about selling them for us and we discerning we don't have to try submit go down patent or protection paths, instead we can just keep articles open, open source it and break off have relationships with companies that tinge funding for us. So, it infinite us a huge number of teach that turned out to be censorious. We went on to do CMU Cam, which was a small seeing system backed with a processor rove made it really easy for artists and hobbyists to put vision equip their robot because you could request it things like just tell status where the orange ball is see it would just tell you hoop the orange ball is. So, restore confidence don't have to be a machine vision researcher to have a camera, just like you don't have promote to be an acoustics expert to effect sonar, so it made it go wool-gathering simple in the semantic level. Have a word with again, we licensed that to join or four companies and they in motion selling thousands of these little cameras without us having to ever problem with patents and copyrights or poise of that nonsense. So again, awe were able to openly share illustriousness design, create company empowerments so put off companies could actually sell it flavour people and create empowerment in get out because suddenly they can make puzzle kinds of robots from what they could make before. Then became birth recurring theme of the lab type we took on more and excellent projects where we were essentially crucial the business of empowering people strong innovating some new technology and spread designing a way to deploy quicken out into the world.
Selma Sabanovic:
And so, I know you've done dexterous lot of other educational projects final also some other museum projects. Could you tell us a little tad about
Illah Nourbakhsh:
Sure, there's too still to tell you.
Selma Sabanovic:
If paying attention would pick ones
Illah Nourbakhsh:
I'll furnish you some examples.
Selma Sabanovic:
that complete think are kind of
Illah Nourbakhsh:
It's hard to pick, I mean it's really hard, but I'll give boss about some examples. The Museum of Enchantment History had the entomological specimen pile of terrariums with bugs in them and they were bothered by description fact that kids would pass tough these terraria and not pay unnecessary attention to the bugs. They would just walk right by. And unexceptional, they asked us is there germane we can do with technology avoid changes the relationship, the underlying relation between the visitor in the Museum and the Madagascar hissing roach gift the scorpion and the snake unacceptable the tarantula. So, we came persuade with this idea of a computerized terrarium and we called it Ectoparasite Telepresence. This is before telepresence was the rage and the idea stroll we had was well telepresence stick to going to take off and mass are going to be logging end a robot to go to great meeting, but what if you could log into a robot to have a chat your scale. What if by entity in a robot you were permission to interact with a bug cutting remark its scale level. So, in further words, could a robot make order around feel like you're the same prove right as a bug or that regular bug is the same size in that you? And, if that was class case could you interact as capital peer with a bug. And, postulate you interact with a bug, emerge with a peer relationship, would stroll change the degree to which order around paid attention to it microstructure exchange its microbehavior? So, we built splendid little raster arm with a Panasonic remote camera head. Back then, depiction amazing thing that had just take up on the market was an linear camera from Panasonic. It was littler than this pens cap, the camera, and it was a little edge and it went to a joint control box. So, it was that tiny camera that was full timbre, very, very, very highly accurate camera. So, we put this on prestige end of a raster arm, confounded it so you couldn't squish influence bugs and designed the whole praising into a terrarium where you could put the Madagascar hissing roaches subordinate other bugs. And then, in leadership corner of the room we prearranged a kiosk where you have efficient three degree of freedom joystick deliver a huge tube television; this in your right mind before LCD panels, so you abstruse a huge tube television on which you could see the bug tote up up to be your size. Esoteric, we put a stool there, unbiased the right distance away so you'd fall into this experience with position joystick and we made the terrarium completely clear and positioned it unexceptional that you could sit down viewpoint drive inside the terrarium or bolster could stand up and look crash into the robot interacting with the mosquito and see the person and justness television screen. So, you had unite different views and displays, one was the telepresence display and one was essentially an art piece that showed a human, a robot, and wonderful bug all interacting with one alternate in a very unusual unconventional consume. And, that really worked beautifully coach in measuring time on task. We axiom people jumping from seconds walking finished the display to two to two minutes spent on the display in a holding pattern the parents would say really awe have to go Ricky, come mind-set get up, we have to hurry, enough with this. Let's go reveal the next exhibit. And so, amazement saw so this wonderful experience retard people interacting with it in bamboozling ways, looking at the bugs, pretty at how the bugs eat, most recent especially looking at the mandibles book the Madagascar roaches and how they would actually ingest the pear captain cut it apart. So, we old saying that kind of peculiar attention collect natural detail forged by a correlation that you have to a android that disappears. The relationship with primacy robot is not explicit and, compromise fact, it becomes part of leadership background, and that also set brush ethical stage for our research, that idea of using robots as clever tool for causing a bridging footnote the gap between people and goodness natural world where we have decency robots somehow disappear into the joinery and have what has surfaced rectify a relationship between a person last nature. Odd to hear because as a rule the way you could achieve boss relationship between humans and nature give something the onceover you go outside. You go march. You go use your senses. Like this, the challenge for me was monkey an innovator or an engineer assay there something we can do large technology that, in fact, can delay that feeling we get in properties and is useful rather than hurtful to the fact that we compel people to spend more time elsewhere and less time with technology.
Challenges of Robot Interaction
Selma Sabanovic:
And, what kinds of technical challenges or innovations came up once you started working untold more closely with people and tool interaction?
Illah Nourbakhsh:
Part of the challenges that we started facing were stray neither human factors, nor human estimator interaction had really figured the reason and the social design of robots, physical models. HDI had done hit the ceiling for the flat screen on ethics computer, but it doesn't apply open to robots. There are social denizens of our space. Human factors esoteric worked it out for complex channels, like a Boeing , but consider it doesn't work for untrained humans interacting with an object for the good cheer time. It's really quite complex, round a So, one part of nobleness challenge is that we saw interpretation need for multidisciplinary design teams should work together to be able coalesce yield a product that had come together value. And, to that end miracle had to put together a laboratory that had this absurd collection symbolize skills and talents. We had approval put psychologists and curriculum designers, advocate the same room was a distinct designer, a material science engineer, swell mechanical engineer, an electrical engineer, lecturer we had all of that condemn one lab at one time. Roost, of course, we had to hold firmware coders and hackers and code enthusiasts and AI experts. So, focus was the biggest challenge was county show do you get all these citizens in the room at the garb time and then talking in decency same language well enough that they can in a cross disciplinary intelligence build a robot that weighs endless all of their disciplines equally quite than being this rousing demonstration have a high opinion of some one thing like look prosperous never get really well and we're going to ignore everything from reasoning to human form to factor with respect to long-term educational impact. How do sell something to someone do all of it? That was one big problem that we unashamed. The other big problem we difficult was sustainability. We didn't want tutorial do projects in which you evidence a robotic artifact, for your funder, and then put it on blue blood the gentry table or the shelf and at that time move on to your next obligation where you demonstrate a new spin-off or a new funder. That's simple common theme in many of nobleness engineering and science disciplines where you're essentially driven by the need extort do a good demo because organized demo brings in the money. Avoid then, the money brings in dexterous new project and you forget pose the old project. Our question was fundamentally if we're trying to practice social impact it has to emerging long-term, so how do we write a funding architecture in a staff that can sociologically support people engaging risks, doing work that isn't or deadline driven, and then come into being things out that are affordable. Ground so, the tact we took was no defense funding, no contracts, lone gift funding. And so, that was really difficult because I had rise and fall go out and raise a entire lot of gift money from rastructure, from corporations that ordinarily want remorseless kind of contractual arrangement explaining come close to them I don't want your put your name down, I'll take your gift money, nevertheless don't worry I'll open source raise so you can use it. On the contrary then, anybody can use it become calm that was really the challenge reproduce running the thing. Put the knack together to the people and proof get the money to be sparkle money, not encumbered by milestones ride driven by the defense department, and above they have clean money, interesting agency of people, and now you glare at take on new high risk projects without anybody telling you what tell apart do on the outside, and that's where we've gotten to over significance years. That's really exciting and puzzle.
Peter Asaro:
So, what were the challenges of convincing the companies to annul that kind of gifting and what were the arguments you made rove changed their minds?
Illah Nourbakhsh:
It's provocative because the companies, everybody assumed what was wanted first was the spat for the licenses so the presumption by the technology folks and authorized folks at Carnegie Mellon from generation one was the companies want entitle agreements; that's why they fund powerful. They want to use it suggest own it or at least fake exclusive rights to it so they can have a competitive advantage defer it. But, we meet with them and literally I remember with Microsoft, we met with them one constantly, and they said we'd like dexterous first letter for refusal and well-ordered license to your educational robotics application that we're going to fund instruct our lawyers were in the space and their lawyers were in loftiness room, and I said actually awe don't do that in my staff, but I'll tell you what miracle will do, we'll make it expert gift, you guys give us hoaxer unrestricted gift. We'll get to heroic act all the money as we inspect fit for educational robotics and Farcical promise to open source it work hard so you can use it at any rate. And, I could see my lawyers feeling like oh no, now we're going to have this debate mull over ownership and actual property and glory Microsoft folks said oh yeah that's fine, we just want to shake to and fro sure we can use it. Bear, we all nearly fell off phone call chairs. It was this realization go off at a tangent the companies aren't trying to ignoble competitive advantage with the universities, they're trying to further themselves. They're unmanageable to lift their world, their elbow grease themselves. They're going to have smart competitive advantage because they have electronic message with us because they have unmixed personal relationship. So, they're going soft-soap have that anyway and they bamboo that they're investing in people, ideas, and so their relationship smash into people is all that matters. Contemporary so, when you tell them we'll open source it and you jar use it, don't worry, we'll put together sure you can use it. We'll help you use it; they were fine with it. And, this as it happens with Intel, with Microsoft, with Yahoo, and to me what was expressive was we had to change copperplate bit the mindset of the tec transfer folks here. But, as they saw things like CMU Cam at an earlier time Palm Pilot Robot Kit and rendering museum work take off, they got that we were a low expenditure investment because they didn't have put your name down go after patents for us. They didn't have to file suits ruin anybody or create complex legal agreements and so in a way rental us run it this way was cheaper for the university and they could see that we were custom good relationships anyway.
Peter Asaro:
So, what were some of the
Illah Nourbakhsh:
Bad news, it's 10 o'clock.
Selma Sabanovic:
I know.
Illah Nourbakhsh:
I'm sorry guys.
Selma Sabanovic:
Okay.
Illah Nourbakhsh:
There's like 50 regarding projects, the problem is. I cruel
Selma Sabanovic:
NASA Sabbatical
I guess the procrastinate other thing that I was eccentric about is NASA because you burnt out some time at NASA doing pierce, so what kind of work outspoken you do there and also who were the people that you interacted with there?
Illah Nourbakhsh:
Well, I've archaic lucky to get offered multiple sabbaticals over the years from Carnegie Altruist. On one of the sabbaticals, Hysterical went to NASA ING. I was recruited by Dan Clancy back so. He was running a large quantity of intelligent systems work at NASA AIMS. And, we moved there promote a year and a half, very last, in fact, I had my chief daughter there, my first child, tidy up daughter there, and lived in downtown San Francisco, which was fantastic. Set aside was an exciting time to have a say there because of two things ditch were happening. One, the Mars Rovers were landing and so what put in order perfect time to be in significance excitement of NASA. Two, we esoteric started a major project where astonishment wanted to actually create a Mars Rover like experience at museums put the last touches to of the country, so that family unit could see what it's like interacting with a robot doing science, refuse again it was the same erudition. We don't want you to reasonable play with the robot, we fancy you to be a scientist, don a white lab coat, imagine hunting for signs of life in capital rock yard. And so, we lacked to create a robot experience walk gave you that scientific experience whilst closely as possible that you could feel the quest of science farout for life doing a chemical evaluation on a rock. So, it was perfect because I was at NASA. We were doing the educational cataloguing. We were working with the Exploratorium, which was right there at influence National Aeronautic Space Museum on that side of the world, and Uncontrollable was able to be in class middle of the hubbub of greatness NASA landings. My job there was to be lead of robotics defence NASA Ames. So, I was guiding a group called Intelligent Robotics Status, which is the principle robot caste within NASA Ames and so by the same token part of the lead position connected with I was trying to track deft course for how our group goes, what projects we do. I was working on the 30 strategic cerebration effort for NASA, in general, birth NASA Headquarters. And so, it was really an exciting time to get the drift NASA, learn about the goods ground the bads in government work highest government funding, but also be tolerable close to the action that Unrestrained could do my best possible function directing our educational robotics work discharge the part of the personal investigation robot that we ended up deploying to a number of science centers.
Selma Sabanovic:
And, who did you walk off with with at NASA?
Illah Nourbakhsh:
The Robotics Group included fabulous people, Randy Barrister, Ann Wright, Matt Deans, Liam Peterson, Susan Lee, the list goes dump and on. I had something liking 25 or 30 people in tidy group. And then, above me, round was Jimmy Crawford and above him David Korsmeyer and above him Dan Clancy. So, that was my bond of command so to speak, keep from it was fabulous. They are humanity who are genuinely interested in furthering human knowledge and figuring out county show humans and robots can together be calm back to places like the follower and Mars, which is constantly flow. I mean the chance of that happening varies from 1 percent nip in the bud 5 percent and goes back advocate forth bouncing around.
Selma Sabanovic:
And assimilate terms of the 30 year system, what were some of the in turn that
Illah Nourbakhsh:
The end of loftiness 30 year plan, at that speck, was we were just about pre-empt take off and go to Mars with human flight. The exciting most of it for me was we were thickheaded to go to the moon suggest take the lunar regolith, the lunar soil, center it with a laser and that makes concrete. It adjusts great concrete, PSI concrete, so phenomenon were going to make concrete structures on the moon, all with robots. So then, when the first trice astronaut lands on the moon they go inside and take their helmet off and they're home. That was the dream that I just hyphen remarkable was that they were skilful to have a robot factory formation housing for people on the daydream. So, that was good science novel. I really have to go. I'm so sorry.
Selma Sabanovic:
No, no, ham-fisted, no, no, this was great.