September 20th, 2012

Transcript – Tilden

Interview with Mark Tilden,


ROBOTS(Per): Today we are here with Mike Tilden and we are going to talk about robotics in general, the progress we have seen over the last couple of years, or basically anything he would like to share with us, so we are going to start looking at his robots and go through them a bit and we will discover many interesting things I think.

Mark Tilden: Okay.

ROBOTS: So should we start anywhere that you think is significant? We talked to you in 2008 when the Femisapien had just emerged in the market…

Mark Tilden: When the Femisapien just came out…

ROBOTS: Yeah, should we pick up there and see what happens or…?

Mark Tilden: I’m not sure if we must talk about the history that came before Femisapien.

ROBOTS: Yeah, let us do that…

Mark Tilden: What was interesting is that Robosapien was still very much of its heyday, and what happened is that I had taken my BEAM technology and turned it into this inexpensive humanoid technology, and then through the WowWee company we wound up putting [out] over a dozen robots, which were all just advanced experiments, to prove my idea of optimizing performance-to-silicon, that is try to come up with something which is as cool as any humanoid robot you can manage for just the smallest amount of money, because I believe that we are not going to have a robot uprising until the robots are affordable, and that is a big thing about a lot of my bottom-up technology up here. Femisapien was at the time probably the most sophisticated robot humanoid I had ever built. It had been five years since we built a Robosapien, which was like a walking gorilla, and the fact was that she still is a very efficient, very human looking humanoid, which is interesting because she looks human on the outside, but she is all, you know like advanced, evolved robot from the inside. And since then we came up with Joebot, which can actually listen and talk back to you in English, although he also speaks German, Dutch, French and Chinese. And then, of course, we have many other robots. The business has faded away, that is the market for entertainment robots is essentially saturated now, and what I am currently doing now is exploring various different spaces, in trying to make robots that don’t just entertain you but actually do something real, something that you actually need to get done. That is going to be an interesting thing for the twentyfirst century, because it has to be and look as sophisticated as a cell phone but as easy to direct with as an R2D2, and I have got some beautiful designs along those lines, but we will have to see if they will actually make it to the market at all.

ROBOTS: What kind of tasks are you addressing?

Mark Tilden: Here is a simple thing. You are away from your house for, say, up to a month or so. Would it not be nice to have something that walks around, and every time there is a noise all of a sudden it takes a photograph, and if that noise continues it starts taking video – okay? – and then ships that to your cell phone or e-mail account, no matter where you are, so even if the robot is destroyed, the fact is that the information, the evidence, is still basically there in your pocket. Is your brother coming in and stealing your Halloween candy? That is like the low-end application of that, but the top-end application is something where you have a hyper-efficient robot. And that is one of the nice things about the biomorphic robotics technology that I have developed; people do not recognize just how efficient it is when you basically build something that is structurally, mechanically and electrically as efficient as they are, you wind up getting battery lives that could be measured in man weeks, and not just man minutes, like you get with a lot of these what I call force-forward, forced feedback control technologies, like these humanoids that you see out of Japan. The maximum battery life is maybe an hour. Okay? Well, if the robots ever try to take over, that is going to make for a really short war, but if you need somebody to actually do work for you, you are going to need at least three weeks of it wandering your place without you having to change its batteries. These are the sort of realities that [?] going to apply and put our technology into.

ROBOTS: As somebody that manages properties, I certainly see a use of something that is more flexible. Oh, certainly I could put in a system doing one thing, but I want something flexible that could adapt to whatever task was at hand. As you said there was a sound, there was a problem, I get information and then I can make the robot take basic, simple actions that would be just great.

Mark Tilden: Well, simple things, like something that you could walk up and down behind the walls of a building, walking through the pipes, walking through the electricals, looking for electrical hotspots. Were you silly enough to put [in] aluminum wiring for example? Woohoo! Then you could be having a hotspot problem, and of course even with older properties being able to put these sort of things, and knowing that they have the ability and versatility to be able to give you information, accurately without the necessary costs that would come from [?] putting in a totally integrated system, and wiring a building is a massive cost. But if for exactly one tenth the cost you could do a small high-reliability robot, that will run for years, then, basically, you are looking at something that is pretty cool. These are the range of applications. I have speculated that the future robotics is going to be robo-tourism, so in other words you could be wherever you are on your laptop and then run a robot in, say, Australia to look at resorts; robo-shopping, where you can be in Switzerland and you can get a robot to go shopping for you in Hong Kong – right? – four or five dollars a minute, and then ultimately from what we learn from these things, we could finally get to the holy grail which is a robo-home-assistant, that is a thousand dollar humanoid that will look after your parents, in their own house, for as long as they want to stay there, and that is something which is fairly important. It also has to be built out of rubber, so if her mother beats the crap out of it, it will not hurt her hand; it will just go away into the closet.

ROBOTS: Safety and security, we touched on that a bit, is also very important. Do you see any problems we have not solved and any roadways ahead there..?

Mark Tilden: Oh, yes, huge numbers! For example, ones I am currently working on right now is robot self-grooming. Human beings basically slough off our skin, and we can take a shower. Well, how does a robot stay clean, especially if it is in continuous operation? It is going to get dusty, it is going to get dirty, it is going to get filthy, and the thing is that it needs to have the same sort of basic skills that you see on a common house fly, where if nobody is trying to kill it, like, “Okay, I am just going to scrape my eyes off and clean my body.” We are trying to come up with robots who can do that. One of the big problems we are facing right now is recognition, avoidance, and path navigation, so once you have got a human-sized robot that can walk as fast as a human being, first off it scares people, but second off how do you solve the young child problem? That is it is one thing to walk through a mall, but if there is a small child they do not have interactive operations that you would normally expect from grownups. All of a sudden you have got a machine, that is walking up to a child that does not know to get out of the way of a large walking device, and how you avoid these sort of traps. We have had to build roads around the world to separate cars and human beings, but how do robots do this? Robots are supposed to work both indoors and outdoors and even behind walls.

ROBOTS: …and together with us, in another sense, rather than in the car, we are in the car and we can separate ourselves from, with the road being one thing and the rest of the world being another thing, but robots should work with us, [in] much more complicated interaction?

Mark Tilden: It is exactly right, my colleague, R.I.P., Brosl Hasslacher and I came up with a concept called Parallel Life. So in other words, there is us and there is nature, and rather than robots being our agents for forcing nature, what they should be is something which is sort of like a parallel life between that. I wrote stories talking about my old robot robo-ecology, and some of them are over there, where the robot venous fly trap would crush and kill flies that will drop flies onto the floor, where my robot dustbunny cowboys would sweep them up, and the robot window washers would scrape off all remnants. And it is amazing because all these things were self-contained, solar powered, dishwasher safe, and that was beautiful, all of a sudden realizing that this was something that was small and elegant, so even now, even though you can hear some little robots still living in the trap behind the little glass containers here, the cool thing is is that I found is if you are going to have something, a robot that you can live with, it has to do a task but it has to do it quietly, elegantly, and of course in conjunction with human beings, not just under control of them, so this is why I distinguish autonomous robots from, say, teleoperated robots. It is one thing to take a cell phone and drive a car around, that is remote control. This is something which essentially is a whole class of autonomous, self-motivated, self-actuated vehicles which are capable of doing projects and tasks for you, even when you are not around to supervise them.

ROBOTS: That would be very helpful I think, and also from a security point of view it would be nice to know that everything is okay and…

Mark Tilden: Sure, there are little things, for example if you do a lot of traveling around Asia, like I do, one of the things that I have put out is a silicon modification of my little friend, the Roboboa, over here, which basically is a big noisy snake-like machine, but imagine a small something. So you are sitting on a very sketchy, very strange bed in a hotel somewhere, and you go to sleep, and all of a sudden your alarm clock wakes up, scans the room, so if anybody tries to come in it makes a noise loud enough to bring anybody to come running. I mean these are the sort of simple technologies that will become very [visible?]. But it cannot just be that, it also has to be a good alarm clock; it would also help if it was a cell phone charger; it would also help if… So, in other words, what would it take to come up with the universal R2D2s that we have dreamed about for a long time, from Star Wars? Remember, R2D2 was a great fictional comic device, because he always came up with a tool for whatever he needed, right, even in the last movies he was able to fly. Well, why was he not able to do that in the first movies, when he was stuck in a swamp, right? The thing was it was great for the science fiction movies but the thing is what would it take so that you can have a technology that even your wife would want to take on a travel trip, and it would perform actions and do things for her that essentially are necessary?

ROBOTS: Basically a technical companion to us like a cell phone is a communication companion but this is a companion that could lift stuff for you.

Mark Tilden: A passive companion. And what is interesting is that Robosapien was a very interesting social experiment from a lot of points of view. In 2004 Robosapien hit the markets, and the fact was that children around the world saw their first interactive capable robot that was their size. Now they are a little older, and now these robots have become almost like a passive background for the entertainment industry. The market was there, it sort of faded about. Cell phones, we are still paying too much attention to them, but after a while I think they are going to become passive as well, we will not even care how they do what they do, they just basically are going to be able to do it for us. What I would like to do, we are still waiting for the next generation of robotics to best get into that. And this is what we did with the field of robots; we put out 22 million of these entertainment robots and the great thing was that basically told a lot of people what can be done, what you can expect from an entertainment robot. We saturated the field, started with Sony and their little Aibo dog, [and] basically ended things like the Joebot, which is able to walk, talk, and listen for less than 50 bucks, and that was great. But now everybody is going to be asking “Okay, great, well it entertains my kid, what can it do for me?” I am a grownup, you know, like I am a teenager who basically has his own cell phone. This is why we have done things like explorations on, for example, the Sphero, on the robotics technology that is coming out there, that is trying to meld the passive technology of the app world with the realistic expectations of what we can do in the robot world. No wife is going to let you spend three or four thousand dollars on a toy robot for home, while the kid still needs clothes. These are the sort of problems that we are trying to face, but the cool thing about the BEAM and biomorphic technologies is that we are able to reduce the costs of building something for real and put it on the shelves for dozens or hundreds of dollars rather than thousands, and we are basically trying to force that technology now from the entertainment to the real. We basically succeeded in our test project.

ROBOTS: Which one?

Mark Tilden: Our toy project. All of my robots up until now. When I was a scientist I was developing the technology, here in HONG KONG we basically took the technology, applied it to the market, and it wound up working, and what we are doing is taking up that technology now to a brand new level. That is that it is all social experiments, okay? The reason why people look at the old science fiction movies from the 1950’s and like “Oh, the evil robot will destroy us all.” Well, because evil robots were not around, I mean even good robots were not around. They were not common enough, but now what we have done is we basically fulfilled the social contract of letting people know what robots can actually be and how scary they actually are. And one of the things that I’m glad Robosapien did, as with all of our robots, is they have shown an entire generation of kids that robots do not have to be scary, that even these guys who are as efficient and as powerful as we can possibly make them are by no means a necessary threat, that basically just like the same way you are always smarter than a dog, even though the dog might be sharper and faster and stronger than you are, the fact is it is dumber and this is a nice thing at least right now is that robots are dumber, and we can actually start using them, and basically [be] looking at a brand new level of marketing. It is a social experiment taking it to the next logical level. How do we make robots a passive part of your life that your wife will say “Oh, hi honey, while you are by the store, pick up a couple of robots will you?” – you know, that sort of thing?

ROBOTS: Or simply like today with cell phones and TVs, for instance, stuff around us like that, that we have become dependent on them. We use them; we have a deep need for them. We cannot just say that this is a fun thing on the side, but is an integral part of our life. It does critical stuff for us and it also does that in a general way. For instance, I am thinking if we have one of these smart robots in the home, we could check, did I turn off the stove, I forgot to take the laundry out of the washing machine, put it in the tumbler. Certainly we could have single devices that we could connect our stove to the internet and stuff like that but it wouldn’t be general, it would be one system for every task rather than one system…

Mark Tilden: Exactly, but the thing was is that we are not talking about something [that would appeal to us], we basically love these sort of things, like the nature of technology, and we will basically embrace it; we basically learn the necessary background. But, if you were to try and put someone like that into, say, your grandmother’s house, she does not want a new internet connected stove, she wants a machine that basically will, when she is not around, come out, clean up the kitchen, turn off the stove, do the laundry and then put itself away, so she does not have to see it. She wants a mode which is sort of like “I see you, I do not see you.” And that is not just making something which is compatible with the customers of the future but the customers of the past. This is one of the big things, better robots for better humanity, and that is not making something which basically comes with an instruction manual and you have to [teach] somebody a brand new technology just from the get go. What you need is something, which basically fulfills an awful lot of the social contracts that I remember from old sci-fi. If you read a lot of the original Isaac Asimov, like from the 1930’s for christ sakes, you realize that there were things like robots that look after children and they could read books but they were not able to talk. There were all kinds of little conditions, but the thing is that these were social interactions, moral choices, nothing more than morality plays, for what you would do when all of a sudden you disposable human beings. That winds up having its own horrible ramifications in and of itself. But the thing was that we can put up with this in the same way that you can put up with, say, dogs and cats and sheep. Yes they are cute and fluffy and stuff like that but they are dumb, so we do not feel too bad about turning them into dinner. What is it going to mean, for, say, real life documentaries, when you can finally have a humanoid camera man that for money you can give the camera and say “Okay, I want you to swing on that vine over that live lava.” There is a new volcano exploding into Iceland. “Go in there and film it from close-up” – right? – because the robot camera man is not as worth as much as incredible videos and the information that you will get for being that close. You have got to take a look at the horrible reality of the way that human beings are living, like sending people into space is getting more expensive all the time.

ROBOTS: So it is dangerous…

Mark Tilden: This is just it, it is exactly as predicted by a gentleman by the name of [Dockerty?], back in the 1990’s, and he said that by 2012 the cost of insuring an astronaut will basically be the same cost as sending the astronaut into space. That happened, and as a consequence now I have got an awful lot of my fellow colleagues in the United States, like astronauts, just sitting around in their space suits watching TV, because there is no place for them to go.

ROBOTS: We could pay for the trip but we cannot insure them.

Mark Tilden: But there is another way, and that is how could we reduce the cost and the risk of going to space? Well, that is if all of a sudden you have robots that basically people the planet. The thing is that when we were talking about parallel life, what we were doing was we were talking about putting in a parallel ecology, the technology necessary so you can surround yourself with machines that automatically replace the things that mother nature does when you are out in the jungle. Where is that pertinent? Well not here much on earth, because we have got nature, but on the moon for example? You cannot go to the moon because it is dusty, the dust up there will [?] clog machines very badly. But, if you were to do something like, say, have giant machines up there so that could go up there, burn some tiles, turn it into a giant beautiful patio, and then you send up new robots, which will sweep the dust off that patio, when you go to land on that patio, all of a sudden you are living in a dust-free environment. All of a sudden you set up your first mall in space. The thing is, is that those little automatic machines, and that is well within our technology, would be the necessary step to not just colonize on the moon but to colonizing pretty much any place in the universe. For example these guys who are going out to get asteroids, they just have to take a look at some of the Juno missions, and they realize that the real problems with asteroids is that they’re mostly extremely fine, abrasive dust, and if you disturb them all of a sudden you are living in silt, and you have to wait like months for that stuff to settle down even in the microgravity of space, so what you need is something out there, [basically you want an AI?], “This asteroid is dirty, robot clean it up.” And, once you do that, once you have actually reduced these things down so that you can live there, you suddenly realize that this is what the future is really going to be, not just robots that are doing the science fiction things we want, like the social interaction, but things that do automatic processors for us, so it allows us to live in space without having to worry about “Oh, did I hook up the appropriate air hose?” – something like everything from an intelligence space suit, but the idea of terraforming or sheet-forming an entire surface of the moon. Not just building tiles, but making sure that there are low efficiency solar cells, all connected together, so all of a sudden you land and you’ve got not just a clean environment but also power. Well hey, that is a lot less material to have to take to space. But also it is not just there, for example the desert, in places on the planet earth. Let us face it, we’re growing more and more and desert takes up an awful lot of space. Well, fortunately there is a lot of sand there, and sand makes two things: bricks and solar cells.

ROBOTS: They will be the same, small, [?] capability intelligent machines.

Mark Tilden: There are deserts around the world, which are so hot and dry and vicious that no living thing can survive. Not true. Solar powered, self-contained, autonomous robots can survive there very nicely, and if you build them strong enough and well enough so they last for twenty years, and they are made from the same material that they are made from in the environment, then when they break down they are just nothing more than part of the natural world. I like these ideas, I like the idea of things like botcrete, where you just have these little hexagonal hextiles, you just drop them on top of the roof of a building or tent or wherever you are and they self-assemble themselves into a little self-repairing, self-assembled solar cell and all of a sudden you have power to run your own refrigerator. I mean that is what is so cool, and when you want to clean up your solar cells you just pick them all up and put them into a bag. Solar cells right now, they are great and it is a future of technology, but the fact is they are large and fragile and are hard to manage, but if you had intelligent individual solar cells they sort themselves out; you bust them up and they pull themselves together, and when you do not need them you just put them in a bag and you can carry them like you would a sack. So wherever you go all of a sudden now you have [?], you want solar power to charge your cell phone, or run your refrigerator, or run anything else, you just drop the bag; they pull out; they pull themselves into a little solar powered mat, and you have all the juice you need in the world. And, if you lose a couple or dozen, who cares?

ROBOTS: Yeah, because since they are identical and made in large quantities and really small, each individual one does not cost that much.

Mark Tilden: It sounds sci-fi, except that the prototypes that are right behind you, and they have been working now for twenty years. They are the future here, boys. But the thing what is interesting is take to a look at the way that is different from the conventional, like top-down, robotic approach, which is everybody wants a C3PO, or everybody wants an R2D2, and it is the assumption that we have to come up with the singularity; we have to find some way to put a Watson a chip, and then they basically work. Well, it has not worked that I have seen. What is frustrating is that the current trend is still trying to build better brains, whereas I have been pushing for years, you [need] better bodies. Once you have got really good bodies, then you can start putting in better brains and the bodies will match that, but good programming will never make up for bad mechanics, and that is one of the things we are trying to show here. The great thing about the Robosapien line is that it is now getting old enough so that when you pick these things up, say at a garage sale, the cool thing is that it is now cheap enough for you to go in and hack it and modify it and improve it, and hopefully this is not just going to generate a whole new generation of robots, but roboticists. And that was the big reason that I went into toys and kits, and a lot of these things are still online and available through and eBay, and as I said just check out your local junk shops if you want to get yourself a nice little Robosapien, who is able to still burp and fart his way into your heart.

ROBOTS: And you could disassemble that, because it was not very expensive to buy, if you do so some error you just go and buy another one, and you could build whatever you want and whatever you could imagine with these parts.

Mark Tilden: These are designed to be hacked in exactly that sort of way. I mean, for example, chances are in your house there is probably a junk drawer filled with extremely obsolete but too expensive to take apart cameras and cell phones and [KVs?] and cameras and radios and all kinds of stuff. And it is just a stigmata, like “Oh, that camera used cost me three hundred dollars; I know its worthless now but you cannot take it apart from its motors.” Well, the cool thing about these things is that one screwdriver takes them right down to the ground, everything is labeled, everything is color coded, and these things [are] just loaded with secrets. And I’d glad to say that there are a lot of websites out there, for example, that still have links and connections to basically getting a lot of these things out, how to actually explore them out, do things with them, and build new generations of them, which was what I was basically hoping for.

ROBOTS: The labeling and the documentation I think is also a huge difference between these kind of systems and the junk camera, because there is no documentation and there is…

Mark Tilden: It is fiddly inside of a camera. Did you ever tear apart a cell phone and suddenly realize “Oh, well, I am not going to repair this right away.”

ROBOTS: No and there is no documentation, even if I wanted to try to, it is all proprietary, it is all locked down, it is all stay out of it, this is open technology although it is closed.

ROBOTS: There are some things, for example I was going to talk to you about, like some of my modular robotics things, but one of the things that a lot of people never quite realized is that Robosapien, the reason he works is because he winds up being based upon a perfect hexagon, so all of his motors and various other things are perfectly labeled, are perfectly oriented to fill these appropriate residence structures, just because of position. We took that same thing, we expanded it and changed it, if you take a look at Femisapien, who is as humanoid as you can get, and you suddenly realize she is just a thinner, taller version, but she turns out to be infinitely more efficient, because we basically improved our efficiencies in a lot of different ways. The point is that one of the things I like to do is take one of these guys out, rip out his digital brains, and then throw in the original analog brain and then guess what: battery life goes up about four times; efficiency goes up about four times; you find all the secret walking modes. What is great about these things is that the body can stay the same but the controller can go very primitive and you can discover all kinds of brand new realms. And do not forget, these guys do not use feedback control systems like they do with standard servo systems, which are expensive. These guys just use the standard open-loop motors, with what is called [kx-squared?] suspension, and they are so tough they can take a tumble down the stairs, so what is great about these things is that as an experimental platform, as long as your electronics it can take it, I know the body can.

ROBOTS: That also gives you the freedom to experiment and to create things because you know you are not going to break it.

Mark Tilden: To make mistakes

ROBOTS: To make mistakes, yeah.

Mark Tilden: Because the fact is, you go to every robot lab in the world, right, and you will find, if they have a beautiful robot, the chances are that it has its own private room, stand back, do not touch, basically be very careful, chances of you getting your fingers cut off, and it is true. You can go to a [?] right now and get these amazing servos, which are basically strong enough to cut a chicken bone right in half, and I said to myself “If that was my finger, Ow!”

ROBOTS: That means you really reduce the value of the technology because you cannot interact with it, you cannot fiddle around…

Mark Tilden: My whole life goal has been trying to make robotics better, and one of the big problems of the top down technique is that it has never broken through that price performance barrier, that essentially is so vital as the bottom-up approach has shown. Crossing over, it is not popular. It’s a lot easier to just sit around and code something. The new discovery of things like the 3D printers is I hope going to change an awful lot of that, provided they find a way to put PCB’s and conductive mechanisms inside them. What we have to do is not just build mechanisms but also electrical mechanisms. The servos I mentioned are very nice, but they are also 700 American dollars each, and you suddenly realize, [woohoo!], you can buy an entire army of Robosapiens for the same cost as that. Now it is just a matter of seeing how far out customers are able to go. What I call the new gentleman roboticists are able to take what is now a homogeneous field of a large variety of different robots, and then start coming up with things that are not just fun and interesting but also eminently practical and hopefully profitable, that is something you could put on a shelf, that will not impress us but will certainly impress the wives and children.

ROBOTS: And you see that when we talked about security, we talked about the bipedal walkers; we know that there is the DARPA challenge out there for the walkers.

Mark Tilden: Yeah two million dollars. I will tell you right now that does not even pay for my vendor’s research and development for something like that.

ROBOTS: No, and for instance are they going to rescue and fire personnel…

Mark Tilden: It is a good idea. No, the fact is it is true, sending something that can basically run into a building, because even dogs will not go into a burning building and find human beings. The problem is you need something that can run into a building and has the same characteristics and sensory abilities and speed as a dog, but it also has to be as disposable as a dog. This is something which I saw years ago when I was building demining robots. [Standard demining?] will charge you $88,000 each year and he can remove up to 21 mines a day, whereas a robot deminer can remove 2 mines a day and it cost $2,000,000, and the leading cause of death is still the same between one and the other, it is heart attacks due to the stress, but the fact is that accounting-wise human beings are just a hell of a lot cheaper to build than robots, and they are the only thing that can be built with [unskilled] labor, I mean that is really important.

ROBOTS: So, that was it for today. I would like to thank Mark for being part of the podcast, and I hope you enjoyed it.

| More