Touring the Robotics and Mechanisms Laboratory at UCLA


hello welcome UCLA alumni and friends my name is Dennis Hong I’m a professor here at the mechanical aerospace engineering department and the director of RoMeLa the robotics and mechanism of laboratory at UCLA are you ready to see some robots yeah come on in so as you enter you see all the awards that our students won from best paper Awards competition Awards a lot of publicity about a I’m on the cover of Forbes magazine probably the only non rich person on the cover so we really started our robotics work back in 2007 with the DARPA urban challenge urban challenge as a at the time was the most difficult competition about building robotic cars so you press Start and nobody touches anything and the robots actually drive the car obey all the rules of the road so we won third place in the world and that really put us on the map and since then we built a robot called Charlie it’s a humanoid robot it’s considered the United States very first human adult-sized human robot and many many different things and today I want to show you a lot of cool stuff so come on in yeah so when you do robotics research that doesn’t necessarily mean that you build robots but we do build robots because of course it’s fun but one more what an interesting thing is you know you can do robotic research and just simulation and theoretical work but unless you really build it and test it you don’t get to really know all the details so we focus a lot on building actual prototypes so this is the workshop so you can see a lot of tools we build things there’s a new type of robotic hand that we’re working on the secret project know a lot of different things so how do we actually make the robots we have electronics on those kind of things up do you probably know about 3d printing have you done 3d printing before yeah so we do have 3d printers so we call those added Manufacturing which means that we add material to build the part but many times when it builds make metal parts we call it subtractive manufacturing so I’ll show you one Ryan do you think you can run something okay come on in so Ryan is one of our undergraduate researchers we about how about 20-something graduate students and about 15 undergraduate researchers as well so this is a machine called a CNC machine so easy way to think of it is it’s a robot that makes robot parts so for example we put this like big metal block right and we design all the parts using a computer and you send that information to this machine and this machine will cut the parts like this and these kind of parts come out so these we use for the robots so from this big metal block to this right so you call it subtractive manufacturing from here to here and Ryan and I think your show us how it cuts yeah safety first always the Ryan is not controlling it the computer is controlling all the motion and there’s a tool that spins like a drill and it’s cutting all the metal and taking away unused metal to make the part now I think it’s gotta take a lot of time so let’s move that way again be careful there’s a lot of sharp stuff around so so we have a lot of interesting robot approaches the first robot there you see everybody knows the Luskin Center right it’s a five-star hotel on campus on a conference center it’s great and this robot is called L.A.R.A. Luskin Automated Robotic Assistant and this robot is actually permanently at the lobby of the the Luskin Center of course we’re fixing it so it’s here but in probably a month you if you go back to the lobby we’ll see you L.A.R.A. so this is a robotic platform to do many different type of research but it actually it’s a it’s a conceit for example if you have questions you can ask the robot and L.A.R.A. answer questions at this point we still we have a pre-programmed question so for example about me like you’re curious about a robot what is your name my name is Laura the buskin automated robot assistant thank you nice meeting you who created you which is over here now being all UCLA alumni and our friends I’m sure that you want to see Laura doing be a clap you see you see la fight yay that’s the spirit so at this point there’s only a pre-canned questions and answers but you can see it has microphone has cameras and the more we use this it gathers more information and we’re going to add artificial intelligence in the future you can actually talk to the robot and it will give out answers so we’re very exciting about L.A.R.A. and it looks cute too so this L.A.R.A. doesn’t walk as you can see the robot is bolted to the ground but if you want to have robots that do useful work it needs to move it needs to go to places if you want to have a robot to do your dishes take out the trash or do work in that Factory or a disaster relief it needs to go from point A to point B we call that problem mobility probably the robot needs to go from point A to point B how do we do that now yeah we can use wheels your car doesn’t have wheels or legs wheels right because wheels are simple they’re efficient they’re great they’re low-cost but we’re trying to build humanoid robots or leg robots of five here or what’s that walk like us why would you want to do that why do you think we need leg robots if wheels are great why a legs this is serious that’s a great oh that’s a great answer if we want to have robots living with us in this environment this environment is designed by humans for humans your stairs had a certain height this for this design for a human to walk up your door handles certain height for people to open it up so for robots to live move around in this environment we claim that everyone needs to be human shape and size and that’s why we tried to build human size humanoid robots and I’ll show you some of our humanoid robots one over here okay cool right yeah this was called Darwin it stands for dynamic anthropomorphic robot with intelligence it’s very cute this robot as a research and developed with the funds from the National Science Foundation everything is a full open-source robot middle button right yeah suffer okay so this robot is fully autonomous just to show you we programmed it to play a game of soccer he likes the color red and can kick the ball alright right foot there you go again you can do it you can do it kick it kick the ball come on the other side right right foot right foot yeah and if he falls down of course he knows how to stand up back again alright so this is just one example it’s not just a soccer playing robot at the open-source robot so people can use this platform to do robotic research if you have II’s a different type of algorithms and coding can use this robot platform through the experiments isn’t that cool yeah yeah yes Father I think he likes you he likes red color I’m not gonna follow the ball oh I like I like your uh I like your pants thank you all right so this is Darwin thank you very much so these robots are great they’re small body these are not toys they are very expensive right but these small robots cannot do any useful work we need to have big robots to do the difficult things yes question oh yes first of all cost having big robots very expensive very dangerous like these metal beasts when you’re doing things if it falls it’s gonna be dangerous too right and then these kind of things are by P robots they don’t scale very well so smaller robots are much easier to control with two legs but when you make it taller it doesn’t scale very well but we do need to have big robots so what we want to show you over here is called for tactical hazardous operations robot now we have some really strong big robots right this robot is designed and developed for disaster relief situations about what six years ago in Japan the Fukushima Daiichi in the nuclear power plant there was a big accident a lot of people died and even now it’s a big problem so when we have these kind of you know accidents we need to send robots to save people’s lives now people it’s very difficult for workers to go in there to rescue people’s life because because the radiation people gotta go so that’s why you develop this robot so this robot can drive a car climb ladders use power tools open and close valves and all those kind of things so at this for it is autonomous but today we’re gonna do just remote control just to show you I think it’s gonna do some self introduction I think don’t have to do everyone welcome to over there my name is Thor RV tactical hazardous operations robot rapid deployment see you I participated in Vanara Robotics Challenge in Pomona California and then in July I went to China to play in Roma Cup 2015 now I have a five-time world champion Roberto I am grateful to meet you okay Thank You Thor so if you get Thor so just like a human it needs to have sensors to gather information from the outside world so for humans what kind of sensors we have what we use to gain information from the world you use your eyes to see right Thor has a camera he sees using the camera he has microphones to hear right and also you have some senses that you don’t have in your face like for example inside your ears you have the sensor for balance so for this robot we have a thing called an IMU inertial measurement unit and use your senses for balance we has four stroke sensors and all those kind of things this is a special sensor called a laser rangefinder so what it is is it shoots out a laser and if there is an obstacle it hits a bounce back and it measures the time and if you know the speed of time a speed of light I need a better time then you can calculate the distance so it shoots out a laser and scans the environment and it knows a three-dimensional train and geometry in front of it so again if you get oh there’s a box there’s something there’s a car so we use those kind of test sensors now it looks like a human right two legs torso head and arms and hands it’s something different from a human do you see anything different what about the configurator something weird over here yeah that’s right the knees bent backwards why will we do that well this is something that we noticed by doing experiments originally it was the knees was bended forward like this but when we’re walking up stairs it tend to hit a lot of things so he said what if you invert it backwards and hey works perfectly fine and its actually better so that’s that’s the only reason why I bend it backwards all right all right so these humanoids are good but the problem of these humanoid robots are they constantly fall down they’re slow they’re expensive they’re complicated right and you know if you wanna make it walk chemi yeah let me show you walk it yeah it walks really really slow all right let’s make it walk a little bit as you’ll see it walks very slow I’m gonna get you very slow all right but yet what you see over here is that this is state-of-the-art as I mentioned it’s slow Josh is behind it because if it falls you can catch it because these are very unstable so we got a long way to go cool thanks guys all right yes question if you guys ever study like animal movements to decide how your robots are gonna walk or quadrupedal movement versus yes absolutely so what we’re showing right now is humanoid or bipedal locomotion so we tried to study how humans walk we collaborate with other researchers and medical community at sports science and also you see these six cameras over here one two three one two three like these are not security cameras do you know what these special cameras are used for we called it yeah we call these motion capture cameras Oh like in in the movie industry computer graphics you know probably seen actors put these like special balls and when you move it it captured the motion of the actor and then use that for comfort a generated animation we use the same type of cameras to study human motion and robot motion and compare that so yes we do study human motion for humanoid robots as ever I don’t also collaborate other people as well cool yeah so you were talking about the senses yes would there ever be a need for him to have either a sense of smell or a sense of touch like to know that there’s a fire yeah very vague sharp question yes we do need but I guess it depends on the situation depending the application smell is a fantastic sense it’s not for humans that’s not that sensitive but for dogs we still use dogs for sniffing out you know bombs and those kind of things because dogs have such a heightened sense of smell so some researchers are trying to develop new type of sensors like artificial nose and with that’s available yeah that’ll be great touch is actually very very important we also do research and touch we put these we call the tactile sensing on robotic hands so when you grab things or try to pick up things you do need to use your sense of touch so yes we do use skin like sensors as well but thank you so much yeah so humanoid robots are great right we need these robots to walk and climb stairs but they always fall down we have a lot of visitors come to our lab and they see our robots and you’re like wow this is cool but after about five minutes they say professor Hong but how come they only walk this fast how come they always slow then we walk with two feet all the time right so stable so fast but why can’t we do this with robots now I can explain this to a actual adult or a scientist but I had this this philosophy if I cannot explain it to a child then probably that means that I personally don’t truly understand it so I start to do more research and try to think about fundamentally why is it important to make robots walk with two feet and I found something fascinating now check this out so for a robot human about to walk like this one of the big problem is your left leg and the right leg has the distance between them and this is a problem why is that because your leg moves forward and backwards right and because of this distance that creates this unwanted twisting twisting motion and falls down now if you look at a ballerina or doing fencing they always walk sideways because if you walk sideways everything lights up so you don’t have that twisting of moments anymore okay problems of them you make your robots always walk sideways right that’s possible but the problem walking sideways is that you’re not really using your knees so we decided to do this okay if you got to do this let’s make this four and make it walk like this right it’s not a human form but it’s bipedal and it’s stable and thus this robot is born this road is called NABi or non-anthropomorphic biped which means it’s a robot with two legs but it does not look anything like a human right so because of this this is forward and backwards body the reason why I have this face is – cuteness but also before we add this face people thought it always walk sideways no no this is for and backwards so it walks like this can you make it walk okay so it’s much more stable it’s much faster its low-cost and lightweight all right so it’s so lightweight this is all this is all right and it can’t walk stabili so this is a I truly believe this is a breakthrough it’s a way of thinking differently literally out of the box and we solved a lot of problems in a bipedal locomotion right cool neat stuff all right what can you show next can we do A.L.P.H.R.E.D. right so you might ask okay this is great so what’s this way but how does it turn right or it’s stable this way but how do you guarantee stable stability sideways it doesn’t have anything so thus A.L.P.H.R.E.D. was born now this is an early prototype of A.L.P.H.R.E.D. stands for cool right so don’t think of it as a four leg robot think of it as a knobby robot what two but two arms however we call this multimodal locomotion which means that there’s many ways of making it walk for example because there’s four limbs we call these limbs not legs because they can be used for walking but also for manipulation like picking things up we can walk with four four limbs such as this as you can expect it walks like a four legs you know creature right but if you want to make it go really fast you can change the configuration like this and then gallop or run like a horse like a dog don’t make it go too fast we only have that it’s gonna run this is just a little bit see and then we’re not ready to show you today but of course if you fold your arms like this becomes this type of a knobby top bipedal locomotion and in the future we might be able to do like this and do it like a cartwheel kind of motion like this right so this is an early version of the robot cool thanks Josh all right so this is all great now one of the fundamental problems of these type of locomotion robots is that these actuators actuators think of it like muscles of things that make robots move normally we as electric motors with a gear now 99% of the robots that exists today use these type of motors with gears so they’re very accurate the strong but accurate the problem is they’re very stiff and stiffness is good if you want to do like assembly task because accuracy is good but if you want to have leg locomotion you need to have something compliant with spring so it’s spring so most of the human rope walks like like just like a robot right that’s because the actuators use our electric motors that are really stiff with gears but what we really need are actors that has compliance right like we call it artificial muscles like animal muscles so we developed a new type of actor called the bear actuator can we run this – yeah so you’ve seen a de Navi robot so instead of these traditional electric motors with gears now we’re replacing them using these special actuators that we developed in-house we are very excited about this new actuator this will enable you know things that nobody’s seen before this is an early early prototype ok just gonna make it a hop a little bit ok something a little right so hopping is actually very difficult to do in robotics not necessarily because of the jumping part but more important because of the landing part because when it lands if you have just a regular motor with gears when it lands because the impact all the gears shatter but because this actuator is very compliant we can’t do all the stuff what if we can’t have a robot that walks with two leg that never falls down or it’s a robot that cannot fall down wouldn’t that be cool can you have a robot that’s so safe that a baby can crawl in front of it without any danger we have built a robot that just does that how can you do that so we cope with all different type of creative creative ideas one way of coming up creative ideas is that we ask ourselves ridiculous crazy questions to ourselves and then ridiculous answer comes out right and sometimes those answers lead to ingenious ideas so one day we asked ourselves what if we can change the direction of gravity it doesn’t make any sense but we ask those questions and ridiculous answers comes out and we came up with a robot called Baloo is Baloo ready we are about the idea that doesn’t work it’s a robot it’s a helium balloon with two legs this is ridiculous but look at this it walks elegantly it can climb up steps it can climb down stairs it can jump right yay and it’s the world’s safest robot yeah you can shove it or give it a hug if you want to yeah so this is a very new concept nobody’s tried anything like this before right all right yes so there’s many different things sometimes we have a specific application like oh there’s a disaster the nuclear power plant what how can you build robots oh we have a problem I need to come up with solutions sometimes we randomly have some kind of interesting idea for a mechanism which has no purpose or no application right so that idea itself has no purpose but we all I almost always I have an idea notebook always write that on my notebook and sometimes I have these call for proposals for research agencies this time we need these kind of technology I just flip through and suddenly I see a sketch like ah we can use this then those ideas which had no value finally finds an application and that’s how it comes about so it all depends on the situation in cases yes yes please oh cannot see so this robot currently has no sensors so this has no intelligence we’re just remote controlling it because this is a brand new concept we want to test that can walk now but yes it will have eyes sensors just like the robot T.H.O.R. eventually we’ll have a computer to think first of all you can’t blow up another one which means that because these robots are so cheap so low-cost like this one it’s multi-million dollar project if you break this has been promised I’m like this because it’s a low cost so you just do another one so low cost robotics has a lot of value to it right and your question so these type of robots solves a lot of problem it doesn’t fall down it’s fast its low-cost it’s great but it also creates new type of problems to these type of robots won’t be able to carry heavy things there’s a solar can all right so it cannot carry heavy stuff yeah and also this robot is not good for outdoors because in outdoors if the wind blows it flows around right so it has its limitations we see this robot as more of an information device so you can have like a tablet with a screen so locks up to you can do you like a cubed walking kiosk or in that factory if there’s a like a gas leak you can send these robots to take samples and surveillance surveillance so it has its own odd and nifty different type of applications all right so bipedal locomotion humanoid is great but what a biology like in nature yeah there’s you know humans and you know walk it to two feet but there’s different types of locomotion as well like insects have how many legs six legs right so why not accept part of all of us yes glad that you asked hahaha we also working on hex up our robots Shen what come in and show them okay okay let’s okay let’s go down that way look if you could yeah gonna go that way so this robot is called SiL.V.I.A. stands for six legs vehicle with intelligent yeah articulation so it looks like a spider but spider has how many legs eight so this is not a spider it has six legs right it can well it’s omnidirectional which means that there’s no forward and backwards on sideways it can go different directions yes that’s correct yeah yeah and then go up a body can go up and down right down and you can do all all changes it’s good like dancing and bleep we believe this is the strongest hexapod robot in the world so can I put this weight on it if I may so I have a very heavy it’s 20 kilograms right and it can handle very heavy weights okay okay now also this is the world’s very first robot that can climb and brace pin walls and climb can you this might not work but we’ll try it over here so it’s going to go through this wall this opening so you are the very first people to see this live we haven’t shown this to anybody in person besides a research in our lab all right so it’s positioning itself you can do it SiL.V.I.A. oh yeah all six legs are off the ground and can it walk up slowly it is like spider-man okay so tries to maintain balance while walking up again you’re the very first people to see this outside of our lab yeah thank you let’s go all right cool I think we’re done with this let’s show the big one so it’s gonna take a few minutes to set up the large one so meanwhile any other questions you have about our robots or in general anything yes please best Idea that never worked we have so many ideas that never worked the thing that you see so the interesting thing is people only want to see our successful robust but behind these we have like 10 times more failures so these robots I showed you it’s a it’s a multi-million dollar project there’s only one in the world of this kind so these prototypes are very expensive and rare so when the experiments you don’t want to break this right so you do very we do express really uh you know carefully there’s many other labs great labs in the world but our lab RoMeLa at UCLA is different from all the other labs in one sense when I do the experiments I tell the students to make it go faster and try to lift heavier things and I tell the students to try to break the robot because if your robot doesn’t fall that doesn’t break you don’t get to learn anything so failure is not necessary bad thing of course if you if we can avoid failure we do we try to but if it fails we want to learn from them so failure is not a bad thing and that answers your question we have 10 times more failures and all of those failures give us a chance to learn about things and that led to all these successes in our lives the greatest most proud products from RoMeLa is not our technology it’s not a robots it’s not our publications it’s our students I want to have students graduate from RoMeLa UCLA and go out in the world and become engines that will lead in the field and create technologies and to make the world a better place yes please pursuing ideas with no purpose yes that’s so inspirational that I think it’s a message that should get out to people in all industries Oh different works of life but we can all do that yeah yeah that’s that’s it thank you thank you so much that’sthat’s what we do so if you look behind you you see that wall that’s not graffiti so that’s our idea wall so that conference table is where we do brainstorming sessions and we write things and you know you see a lot of bunch of math equation things and you know more than half of them will probably never see that there light and light of day and you know there but we come up with all different type of crazy ideas and some of them lead to breakthroughs right that’s what we’re here yes okay so our lab RoMeLa is for research on education it’s mainly for graduate level research funded research but we also have a lot of undergraduate researchers graduate students they come from all over the world to join our lab undergraduate students they get to have a chance to you know get some hands-on and really learn what research is all about now interestingly our lab is very very well known worldwide so people want to come in life especially undergraduate student if I open my email you see like more than just today I have more than 20 emails from people who want to join our lab now many times students when they want to join are they only see the fancy robots that is oh if I join RoMeLa at UCLA I’ll be able to create and make these robots and they come here and surprisingly many times they don’t get to actually build things but they do math and things and some people find this fascinating this is my thing and they become a graduate students when they graduate some students find that oh this is not always expecting that’s not my cup of tea and they leave and that’s actually not a bad thing because you know it’s important for students to really know what they’re getting into right but if you like robotics this place is more fun than Disneyland wasn’t that cool that robot’s called SiL.V.I.A. except for all the world’s very first robot that can climb like this we also have a larger version of that except for robot it’s called hex and it’s behind you so you probably see this you probably don’t know as a robot okay let’s make it stand up huge whoa so we are developing this robot for the application of demining so after war there’s a lot of dangerous land mines underground right and then many times people step on it and they get injured or killed so we ought to get rid of those landmines are fully autonomously so this is the base platform to do that so why do you think we have six legs versus two legs I’ve been talking about all we need human robots all those kind of things what advantage what six legs have over two two legs yeah balance so if you want to walk in a real rough terrain outdoors it’s much more stable we call that static stability because every time you have you know at least three foot touching on the ground so of course ok can go up body up right and down up and down and also can change this orientation yeah so this area is currently open voice this is where all the the sensors and things were going so our vision is it’s not ready yet but there’s going to be a landmine detecting sensor so twenty four seven three six days a day is gonna be walking on a minefield and from here and the sensor so it’s gonna be searching the ground right and if it finds a landmine it’s gonna be it’s gonna sit on it here there’s gonna be a dome so when it sits on it the landmine is gonna be inside the dome thus it’s now a controlled environment no wind inside of them there’s will be lighting so there’s no shadows and inside the dome there’s also gonna be two robotic arms so when it finds a landmine that sits on it and sends a signal to operator so I’ve been up beep oh there’s a landmine I go to my workstation and I tell you operate control the arms to dismantle the bomb so that’s the idea yeah but what if the robot steps on the landmine what was that a problem it’s a good thing because instead of a human robot explodes who cares we build another one right so if it explodes that’s a good thing so you’ve seen a lot of different shapes and sizes and the famous architect Luis Albarn once said form follows function which means that a shape of an object is dictated by what it’s used for same thing with robots depending on the application the shape and morphology and the size is all different all right let me show you one more robots I don’t know if it’s gonna work again this is also the very first time we’re showing it to is it is it working alright this robot is P.E.B.L. Daniel could you tell us about P.E.B.L. what does P.E.B.L. stand for first of all yeah P.E.B.L. okay so this is just the lower body of a humanoid robot said this actually had a whole human robot this was the early version of T.H.O.R. that you seen over there now after experience we took it apart so that’s the lower body we’re using it for locomotion experiments and this is the upper body separately which we use it for something else we’ll show this in a minute but let’s show the P.E.B.L. first so this is a platform we’re trying to study many different type strategies for making robots walk like human well just thing over here is all brand new it happens the past few days so it might not work which is okay if it doesn’t work we learn from it so let’s lift his foot off the ground okay so these are all loose right so he’s not supporting it so even though he changes his body posture can still bounce on one foot now this is obvious for humans but it’s difficult to do in robotics are you gonna be doing the pushing thing or not okay so that’s just one demonstration of balancing on one foot okay we’re gonna set it to a different mode this time we call this disturbance rejection in other words when robot is walking sometimes there’s a wind gust or it hits to something there’s disturbances right and then if a robot that cannot handle is gonna fall so trying to figure out how to do how to change its posture to get rid of the disturbances so it uses different sensors right alright so there are sensors in the ankles that allow it to detect if something is kind of like twisting and so with those on the robot to tell that well I’m pushing I’m person here but at all it’ll try to go back to its original so normally if he doesn’t have this algorithm when you push it it’s gonna fall for like a structure right so you can actually kind of actually really push it a lot here and there’s a limit we pushed the boundaries a limit right so so this was a earlier version of T.H.O.R. so this is retired and we build a new T.H.O.R. and instead of the you know taking apart we actually took the lower body only and modified it for a new test platform yeah recycled it cool thank you very much Daniel all right so the upper body is something different so you know Netflix Netflix there’s gonna be a new miniseries called Magic for Humans starting august 17th and one of the episode is us so i didn’t know yeah yeah i don’t know if you knew it but I’m also a nearly professional magician myself and a nearly professional chef as well I was on the master chef USA by the way yeah season four don’t look me up it’s kind embarrassing by the way so there’s a famous a magician called Justin Willman that you probably know so he travels around the world that visits different places and he actually came to our lab because he wanted to learn about robotics right I’m not gonna spoil all the storyline but he is a magician and he challenges us can you build a robot that can actually perform magic so in three weeks we challenge him and build a robot and this was M.A.G.I. it’s a robot that performs magic tricks so we are not going to show you everything because you have to see it on TV on Netflix in August but we’re just going to show you a sneak peek of one oh there these are small Darwin minis can you can you can you see anything to do about it yeah research using a vision system here and kind objects in contrast yeah you have good taste it’s a face as a boot is a I think what is it suppose we can put your face oh okay there you go Wow so you’re just show one magic trick simple trick okay fall in the cup abracadabra it’s kind of a fun thing that we’re doing but it’s as for studying manipulation doing something like this is very very difficult for robots so this is a sensor you know the Xbox Kinect sensor it’s actually a Kinect sensor so it tries to identify all just had to do the manual placement test alright so those are all the roads that we created there’s some more and there’s some secret projects that we can we’re not ready to unveil but these are Darwin minis these are we didn’t build this but it’s that based on the robot that we built Darwin this is actually a commercial project can you let some of the kids play with it okay give them you can control them yeah let’s let’s do it on the carpet over here yeah yeah whoa this Gangnam style all right you can’t play with it more but again I would like to thank you again for coming to RoMeLa the rewarding some mechanisms laboratory at UCLA and I want to give a round of applause to all of our hard-working students who

, , ,

Post navigation

13 thoughts on “Touring the Robotics and Mechanisms Laboratory at UCLA

Leave a Reply

Your email address will not be published. Required fields are marked *