Boston Dynamics
has lawful launched a brand contemporary Atlas humanoid robot, replacing the legendary hydraulic Atlas and supposed to be a business product. That is substantial recordsdata from the firm that has spent the final decade building the most dynamic humanoids that the world has ever seen, and while you haven’t be taught our article in regards to the announcement (and seen the video!), you might well well also restful form that straight away.
We’ve had just a few decade of pent-up questions about an all-electric productized version of
Atlas, and we had been lucky ample to talk with
Boston Dynamics CEO Robert Playter to be taught extra in regards to the build this robot got right here from and the design in which it’s going to accomplish business humanoid robots (ultimately) occur.
Robert Playter became the Vice President of Engineering at Boston Dynamics starting in 1994, which I’m heavenly particular became motivate when Boston Dynamics restful supposed to be a modeling and simulation firm moderately than a robotics firm. Playter grew to became the CEO in 2019, helping the firm accomplish the tough transition from R&D to business merchandise with
Self-discipline, Stretch, and now (or very rapidly)
Atlas.
We talked with Playter about what the heck took Boston Dynamics so long to accomplish this robot, what the vision is for Atlas as a product, all that crude flexibility, and what comes next.
Robert Playter on:
- What Took So Long
- The Product Capacity
- A Classic Cause Robot?
- Hydraulic Versus Electrical
- Indecent Differ of Motion
- Atlas’ Head
- Advantages in Commercialization
- What’s Next
IEEE Spectrum: So what’s going on?
Robert Playter: Boston Dynamics has built an all-electric humanoid. It’s our most contemporary expertise of what’s been an virtually 15-year effort in increasing humanoids. We’re going to delivery it as a product, targeting industrial capabilities, logistics, and locations which might well well be grand extra diverse than the build you look Stretch—heavy objects with advanced geometry, doubtlessly in manufacturing form environments. We’ve built our first robot, and we take into consideration that’s truly going to home the bar for the next expertise of capabilities for this entire alternate.
What took you so long?!
Playter: Well, we wanted to persuade ourselves that we knew how to accomplish a humanoid product that might well well take care of a sizable fluctuate of projects—grand extra so than our old generations of robots—alongside side at-tempo bimanual manipulation of the categories of heavy objects with advanced geometry that we achieve a matter to to receive in alternate. We also truly wanted to attract finish the advise cases, so we’ve performed tons of background work on making particular that we look the build we are able to conform to these robots fruitfully in alternate.
We’ve obviously been working on this machine for some time, as we’ve been doing parallel pattern with our legacy Atlas. You’ve doubtlessly seen one of the important important
videos of Atlas fascinating struts around—that’s the technical portion of proving to ourselves that we are able to accomplish this work. And then truly designing a next expertise machine that’s going to be an voice of magnitude better than anything else the world has seen.
“We’re no longer anxious to lawful conceal some whiz-bang tech, and we didn’t truly are searching for to conceal our intent to hasten right here till we had been happy that there might well be a course to a product.”
—Robert Playter, Boston Dynamics
With Self-discipline, it felt like Boston Dynamics developed the product first, with out having a explain advise case in thoughts: you build the robot accessible and let people watch what it became appropriate for. Is your design different with Atlas?
Playter: You’re fully valid. Self-discipline became a expertise searching for to receive a product, and it’s taken time for us to truly resolve out the product market fit that now we have in industrial inspection. However the subject of which have has left us wiser about truly figuring out the target capabilities before you advise you’re going to establish that stuff at scale.
Stretch is extremely different, because it had a transparent target market. Atlas goes to be extra like Stretch, though it’s going to be methodology extra than a single project robot, which is vogue of what Stretch is. Convincing ourselves that we might well well truly generalize with Atlas has taken moderately bit of time. This goes to be our third product in about four years. We’ve learned so grand, and the world is no longer like which have.
Is your vision for Atlas surely one of a typical goal robot?
Playter: It positively wants to be a multi-advise case robot. I feel about that because I don’t ponder there’s very many examples the build a single repetitive project goes to warrant these advanced robots. I also ponder, though, that the colorful topic is that you simply’re going to have to take care of a class of advise cases, and truly making them important for the tip customer. The lesson we’ve learned with both Self-discipline and Stretch is that it’s serious to salvage accessible and truly perceive what makes this robot precious to prospects while making particular you’re building that into your pattern cycle. And while you might well well also delivery that before you’ve even launched the product, then you definately’ll be .
How does pondering of this contemporary Atlas as a product moderately than a compare platform commerce things?
Playter: I ponder the compare that we’ve performed all the design in which via the final 10 or 15 years has been wanted to making a humanoid important within the first quandary. We serious about dynamic balancing and mobility and being able to make a choice something up and restful motivate that mobility—these had been compare issues of the previous that we’ve now figured out how to control and are wanted, I ponder, to doing important work. There’s restful tons of labor to be performed on generality, so that humanoids can gain anybody of a thousand different substances and take care of them in an cheaper methodology. That level of generality hasn’t been proven yet; we ponder there’s promise, and that AI will possible be surely one of the important instruments that helps resolve that. And there’s restful tons of product prototyping and iteration that will approach out before we delivery building big numbers of these things and transport them to prospects.
“This robot will possible be stronger at most of its joints than a person, and even an elite athlete, and might well have a ramification of run that exceeds anything else a person can ever form.”
—Robert Playter, Boston Dynamics
For a extremely long time, it gave the influence of hydraulics had been the most efficient methodology of producing noteworthy dynamic motions for robots like Atlas. Has that now modified?
Playter: We first experimented with that with the delivery of Self-discipline. We had the same difficulty years within the past, and stumbled on that we might well well establish noteworthy lightweight electric motors that had the same vogue of responsiveness and energy, or let’s advise sufficient responsiveness and energy, to truly accomplish that work. We’ve designed an even newer home of truly compact actuators into our electric Atlas, which pack the energy of truly an elite human athlete into these shrimp packages that accomplish an electric humanoid possible for us. So, this robot will possible be stronger at most of its joints than a person, and even an elite athlete, and might well have a ramification of run that exceeds anything else a person can ever form. We’ve also when in contrast the energy of our contemporary electric Atlas to our hydraulic Atlas, and the electric Atlas is stronger.
In the context of Atlas’ fluctuate of run, that introductory video became moderately unhappy to achieve a matter to, which I’m particular became deliberate. Why introduce the contemporary Atlas in that methodology?
Playter: These excessive fluctuate of run actuators are going to enable a special home of actions that within the slay will let the robot be very efficient. Imagine being able to flip around with out having to clutch a bunch of steps to flip all of your physique as a change. The motions we showed [in the video] are ones the build our engineers had been like, “howdy, with these joints, we might well well salvage up like this!” And it lawful wasn’t something we had that truly thought about before. This pliability creates a palette that you simply might well well also accomplish contemporary stuff on, and we’re already having fun with it and we decided we wanted to portion that excitement with the world.
“Everyone will purchase one robot—we learned that with Self-discipline. However they won’t delivery by buying for fleets, and you don’t have a business till you might well well also promote multiple robots to the same customer.”
—Robert Playter, Boston Dynamics
This does seem like a strategy of making Atlas extra efficient, however I’ve heard from different of us working on humanoids that it’s important for robots to transfer in familiar and predictable suggestions for folk to be contented working around them. What’s your standpoint on that?
Playter: I form ponder that folk are going to have to became mindful of our robot; I don’t ponder that methodology limiting yourself to human motions. I feel about that within the slay, in case your robot is stronger or extra flexible, it’ll be succesful to form things that humans can’t form, or don’t are searching for to form.
One of the important explain challenges of making a product important is that you simply’ve obtained to have sufficient productivity to fulfill a customer. Whilst you occur to’re sluggish, that’s exhausting. We learned that with Stretch. We had two generations of Stretch, and the first expertise did no longer have a joint that enable it pivot 180 levels, so it had to ponderously flip around between picking up a field and shedding it off. That became a killer. And so we decided “nope, gotta have that rotational joint.” It lets Stretch be so grand faster and extra efficient. On the tip of the day, that’s what counts. And folks will salvage ancient to it.
What are you able to voice me in regards to the head?
Boston Dynamics CEO Robert Playter stated the head on the contemporary Atlas robot has been designed no longer to imitate the human develop however moderately “to mission something else: a friendly quandary to hasten searching for to raze some working out in regards to the intent of the robot.”Boston Dynamics
Playter: The extinct Atlas did no longer have an articulated head. However having an articulated head affords you a tool that you simply might well well also advise to conceal intent, and there are integrated lights that might well well even be succesful to talk to customers. About a of our common ideas had extra of a [human] head form, however for us they always looked moderately bit threatening or dystopian by some means, and we wanted to salvage far off from that. So we made a extremely purposeful decision in regards to the head form, and our explain intent became for it no longer to be human-like. We’re searching for to mission something else: a friendly quandary to hasten searching for to raze some working out in regards to the intent of the robot.
The accomplish borrows from some friendly shapes that we’d seen within the previous. For instance, there’s the extinct Pixar lamp that everyone fell in adore with decades within the past, and that informed one of the important important accomplish for us.
How form you ponder the decade(s) of ride working on humanoids as effectively as your ride commercializing Self-discipline will abet you by methodology of making Atlas valid into a product?
Playter: That is our third product, and surely one of the important things we’ve learned is that it takes methodology extra than some enthralling expertise to accomplish a product work. That you just might well well also have to have a right advise case, and you might well well presumably have gotten to have right productivity around that advise case that a customer cares about. Everyone will purchase one robot—we learned that with Self-discipline. However they won’t delivery by buying for fleets, and you don’t have a business till you might well well also promote multiple robots to the same customer. And you don’t salvage there with out all this different stuff—the reliability, the carrier, the integration.
After we launched Self-discipline as a product a lot of years within the past, it became truly about reworking the total firm. We had to clutch on all of these contemporary disciplines: manufacturing, carrier, measuring the usual and reliability of our robots and then building systems and instruments to accomplish them step by step better. That transformation is no longer easy, however the indisputable reality that we’ve efficiently navigated via that as an organization methodology that we are able to with out problems bring that mindset and capability home to believe as a firm. Truthfully, that transition takes two or three years to salvage via, so the entire emblem contemporary startup companies accessible who’ve a prototype of a humanoid working—they haven’t even begun that run.
There’s also payment. Constructing something effectively at an cheaper payment so that you simply might well well also promote it at an cheaper payment and within the slay accomplish some money out of it, that’s no longer easy both. And albeit, with out the meat up of
Hyundai which is indubitably a world-class manufacturing educated, it can well be truly tough to form it on our like.
So yeah, we’re grand extra sober about what it takes to be triumphant now. We’re no longer anxious to lawful conceal some whiz-bang tech, and we didn’t truly are searching for to conceal our intent to hasten right here till we had been happy that there might well be a course to a product. And I ponder within the slay, that will gain the day.
What’s going to you be working on within the finish to future, and what’s going to you be succesful to portion?
Playter: We’ll delivery showing extra of the dexterous manipulation on the contemporary Atlas that we’ve already shown on our legacy Atlas. And we’re targeting proof of craftsmanship trying out in factories at Hyundai Motor Community [HMG] as early as next year. HMG is usually smitten by this venture; they are searching for to remodel their manufacturing and so that they give the influence of being Atlas as a giant portion of that, and so we’re going to salvage on that rapidly.
What form you ponder different robotics of us will receive most thrilling in regards to the contemporary Atlas?
Playter: Having a robot with so grand energy and agility packed valid into a moderately tiny and light-weight-weight bundle. I’ve felt honored within the previous that most of these different companies compare themselves to us. They advise, “effectively, the build are we on the Boston Dynamics bar?” I ponder we lawful raised the bar. And that’s within the slay appropriate for the alternate, valid? Folks will hasten, “oh, wow, that’s that you simply might well well also judge!” And albeit, they’ll delivery chasing us as expeditiously as they might be able to—that’s what we’ve seen so far. I ponder it’ll cease up pulling the total alternate forward.
发布者：Dr.Durant，转转请注明出处：https://robotalks.cn/boston-dynamics-robert-playter-on-the-new-atlas/
