The Boston Dynamics Story

am I talking about? Well, a few years ago, there were some videos that came out of a company called Boston Dynamics that went absolutely viral, and those videos showed four legged robots traversing various types of terrain, sometimes while enduring some rather abusive behavior courtesy of some humans, and that kind of served as an introduction for a lot of people to this company, Boston Dynamics. But as it turns out, it had been around a lot longer than

you might have anticipated based on those videos. In fact, it turned twenty five years old in seventeen. So what is the story behind the company and its robots. Well, the founder of Boston Dynamics is Mark rayber. Raber was born in nineteen forty nine and he was a studious Fellow. He attended Northeastern University from nineteen sixty eight to nineteen

seventy three and earned a degree in electrical engineering. He then went on to pursue his PhD at the Massachusetts Institute of Technology better known as m i T, and upon graduating, Dr Rayber got a job at the NASA Jet Propulsion Laboratory as a member of the technacal staff. So you could say that he was a rocket scientist, although really he worked with rocket scientists. His main focus was obviously on robotics, and he worked there for about

three years, leaving in nineteen eighty. His next gig was a pretty just went as well. He was an associate professor at Carnegie Melon University's Computer science department and the Robotics Institute. Now, if you've been listening to tech stuff recently, you may remember I mentioned this particular group in a recent episode about Uber, because Uber rated Carnegie Melon and hired about forty of their robotics experts over to help

them with their efforts in developing autonomous car programs. Of course, by that time Dr Raber was long gone from Carnegie Melon. But I just I mentioned it because it's another link to a recent episode. In nine six, Dr rayber took a position of Professor of Electrical Engineering and Computer Science over at m I T, and there he founded the m I T Leg Lab, which sounds a bit weird,

but stick with me. So this was a lab specifically dedicated to researching and designing legged robots, not leg lamps, as I was first led to believe. It was a very difficult realization to say that it wasn't a major reward. But it turns out building robots that use legs for locomotion it's really really hard. Now why is that? Why is it so difficult to build a robot that uses legs to get around? Well, first, let's consider the wheel. Now,

wheels aren't really found in organisms. There are some microscopic critters that have components that are vaguely wheel and axle like, but they're not really wheels the way are various inventions have wheels. A wheel as a limb is not likely to ever evolve in nature for several reasons. For one, any mutation that would set an organism on the path figuratively speaking to developing wheels would likely at first be

a negative rather than a positive mutation. So in other words, an animal with some sort of pro to wheel limb, something that's maybe like part of a wheel but not a complete wheel, probably wouldn't have very many advantages and perhaps have distinct disadvantages with regards to survival, and thus it would be unlikely that such a mutation would be

passed down to future generations to evolve. So that's one thing, is that the pathway for evolution is complicated, and if you have a mutation that is more likely to get you eaten by predators, probably not going to get passed down to a lot of offspring. For another reason, the rotation of a wheel, if not due to gravity, has to come from some source of torque, that rotational power. You have to be able to provide power to a

wheel to make it turn. So locomotion for multicellular organisms typically comes from muscles, right, we have our muscles that allow us to move our limbs. But there's not really a way you could affix muscles to a wheel and allow it to turn freely indefinitely. You know, just think about how your wrist turns. Like even if you had the ability to turn your wrist completely around with no pain or anything like that, you couldn't do it indefinitely.

Your muscles would continuously twist until eventually they would have to untwist like a rubber band. You couldn't just keep ongoing, so that's another problem. Wheels are almost exclusively in the domain of manufactured objects. Now that being said, they're darn't efficient. They're the easiest way to create locomotion for vehicles. It is also the most energy efficient way to travel relative to speed. So it requires a relatively simple implementation to

create a wheeled robot. You don't have to worry so much about degrees of freedom. You don't have to worry about pivot points, or vector controllers or balance. I mean, as long as you've designed the robot in such a way that's relatively stable on its wheels, balance is not that big of an issue. So if you're talking about the three wheeled or four wheeled robot, balance is pretty easy. Once you get down to two wheels, then things get

a little more tricky. You have to build in other systems in order for it to maintain balance, but in general it's not that hard to do. They also typically cost less to develop and less to build than a legged robot. A simple three wheel robot is stable on its own. If you've got that basic sort of triangle shape, it does not require a lot of high end technology to work, just some motors and then whatever sensors and

processors you need to actually control the robot. Uh. Four wheeled robots are are even more stable than three wheeled robots, and they can excel in applications that a three wheeled robot might find difficult. For example, if you plan for your robot to lift things, you could have a three wheeled robot that would become unbalanced depending upon where you put the load on top of the robot right If it's too far off to one side of its center of gravity, it can make the robots sort of tip over.

Our four wheeled robot tends to be more stable in those cases. You can also design a wheeled robot to operate even if it's been flipped upside down. So if the robot is such that the wheels can make contact, whether it's right side up or upside down, you could of the option of continuously operating that robot even if it were to flip over. In fact, I remember some

remote controlled cars that were designed this way. They were marketed as that you would purposefully drive your car and to say a wall, make it flip over, and then you could keep on driving the car just kind of fun. That's the same basic idea, but wheels are not ideal across all different types of terrain. They work best on tracks if you have them, but if you don't have tracks,

they work best on smooth even ground. Depending upon the wheel design and the torque available, they may be able to travel over semi rough terrain too fairly rough terrain, but they aren't great for everything, and if the ground is really challenging, they can prove to be ineffective. So that challenge could be that the ground is just too rocky or covered in debris, or it's composed of loose elements like sand, and in those situations, legs might be

more effective. Your wheels might otherwise spin needlessly, just completely without motion. You're just turning around and you're digging yourself into a hole. Or they could end up catching on debris and pulling them into the works of the robot, gumming everything up. A legged robot with the right control system or programming can step over obstacles and move smoothly over different types of terrain, robots can use legs to select a precise spot on the ground to place its

weight and then shift to move. A wheel has little choice but to just roll ahead, but a legged robot can place its legs and then shift its weight in a very specific way, which is helpful on uneven ground or going up and down the stairs. And depending upon the design of the robot, you can make it where the way it places its leg down, the way it shifts its weight, it can maintain its relative position over

the ground. So it looks like if you if you were to remove the legs magically from the image, like it's just floating over the landscape. But it's all because the way it's able to place its legs and move its weight around. But legs, as it turns out, are super hard to design. They require joints and points of articulation, and you have to have power to move everything and figure out how you're going to create the force necessary to open and close those various joints, and you have

to have lots of precise directions to work effectively. So you need a very efficient computer and you need really good programming so that it can actually maneuver over these different terrain. A leg needs to have more moving parts than a wheel, with at least three actuators to make the leg more useful than a wheel. You can have fewer than three actuators and have a leg, but it's not going to be more useful than a wheel unless

you have at least three actuators. So a lot of the work in legged robots concentrates on how to move those legs. So, for example, you might use artificial muscles with stuff like electroactive polymers. Now, those are materials made out of long chain molecules that change their shape when they encounter an electric field. So here's an example, and this is just one I pulled off the top of my head, and it's not really indicative of how they

all work, but it's way they at work. Let's say that you build a structure and it's made off of this material, and that structure is elongated, it's extended, but if it encounters an electric field, it contracts, it pulls together. This would make it work in a way similar to the way our muscles work. Now, there's not just that kind of method to power legs. In fact, that's a pretty rare one. You're more likely to find electric actuators

or hydraulic actuators. But that's another challenge is how do you actually get the legs to move. Then there's a ton of work on that processing side to calculate things like balanced momentum, weight distribution, path finding, all of these different things that we take for granted we learn it intuitively,

but that doesn't work with robots necessarily. And there's a long history of studies that happened before robotics were even a thing that would eventually inform the design of legged robots, and some of those studies which went on to help future engineers when they were building out these robot legs for the first time. We're pretty grim, as it turns out. So get ready for a gross story that I'm about

to tell you. One study that the Leg Lab cites as being an important milestone in the development of legged robots comes from an eighteen thirty six journal and it involves dead people. The title of this study was Mechanic derminsch lichen I and a Thomish physiologic unter schung fondap rudin Wilhelm Weber w Edward web Webber. All right, so translated to English because that was a terrible German, I know,

my German's awful kind of Deutsches. It turns out it actually means the mechanics of the human and anatomical physiological examination of the brothers Wilhelm and Edward Webber. Now their work is available to read if you read German. But they found that if you took the leg of a corps and you swung the leg of the corpse, so it acted like a compound pendulum. The swinging motion was very similar to the cadence of a leg with a

live person is walking around. That's cheerful. But as it turns out, that would actually be one of those fundamental studies that would go on to inform people who were trying to build artificial legs and mechanical legs. Most of the studies the lab sites aren't nearly so grim and don't involve nearly as many corpses. So for example, another important contribution came from a guy named Edward Moybridge. Now this was an English photographer who developed stop motion photography.

Moybridge was hired by a fat cat named Leland Stanford. He was a former governor and a railroad tycoon and Stanford had this idea, and he needed evidence to support his idea. His idea was that when a horse is in full gallop, at some point during its gait, it will have all four hoofs off the ground simultaneously, and moy Bridge was essentially contracted to prove that Stanford had

a leg to stand on as it were. So Moybridge invented away to use a series of twelve cameras to take photos with an exposure lasting only a fraction of a second, which at the time it was pretty remarkable.

Most photo photography. Early photography had very long exposure time in order to get enough light to actually create the effect, which is why you have lots of portraits of very dour looking people in early photography, because it meant they had to sit still for several minutes at a time, and you know, if you're trying to hold a smile, pretty soon you hate the world and everything in it, so you typically have these very kind of neutral or

dour expressions in those early photographs. Well, Moybridge wanted to find a way of doing this much faster, so he created this camera system where in a fraction of a second it could expose film to enough light to create

an image. He said, twelve of these cameras in a row, and then he used little trip wires not to trip up the horses, but rather the wires themselves would trip when a horse ran through them, and it would cause the cameras to take these photos in sequence, and then you could look at them and see a horse's gallop in progress. The photos showed that Stanford was in fact correct.

There was a brief moment in the horse's stride in which all four legs were off the ground, and Moybridge's innovation became valuable for scientists who wanted to study animal movement, and later for engineers and roboticists who wished to mimic those movements when they were developing robotic legs. Uh, Moybridge and stop motion animation and stop motion photography, I should say, are really fascinating subjects, and I'm sure I'll cover them

in another episode of tech Stuff at some point. Back to milestones, Well, there are a lot of others that the lab acknowledgeist as being really important moments in the development of robotic legs, and they even include a few of Dr Rabur's contributions to the fields, such as three. When Rayber build a one legged hopping machine, it could keep its own balance and travel at a specified rate. It kind of looks like a UFO on a pogo stick. It's constantly bouncing around. The main trick here was just

getting that balance right. So Rayber was able to create this robot that uses a computer that calculates where that one foot needs to come down in order to keep the overall robot balanced while still traveling in the desired direction. Those calculations had to be done fast enough so that they were complete before the robot would come crashing down

after its last hops. So as it hops up in the air, the computer has to figure out the trajectory, the momentum, all of these different elements that the robot is experiencing in order to compensate with the leg so that it can hit the ground at just the right spot with just the right amount of force to propel the robot up again and continue moving it in whatever direction you wanted to go in. So it's actually pretty complicated, It took a lot of work and it's pretty fascinating

stuff off. That basic principle would then later be adapted and applied into multi legged robots with a computer determining where each leg needs to go, how it needs to make contact with the ground, how much force, all just so that the robot doesn't fall over. Rayber would end up writing a book about his research. In the book has the title legged robots that Balance. Pretty much sums up the whole thing, doesn't it. It sounds like it's

a trivial thing. But again, it was really tricky to do because while we're growing up, most of us, you know, as kids, we learned to walk, and we kind of have an kind of intuitive grasp of what we need to do in order to keep our balance barring any unforeseen circumstances like stepping on something that is on uh it's like uncertain ground without even realizing it, or getting

shoved by someone. We're pretty good at keeping our balance generally speaking, assuming all things are you know, normal, But a robot has to think air quotes, think about how to do this. Intuition isn't an option, at least not without some really advanced machine learning algorithms, which honestly, we're not that far along. In the nineteen eighties, this was all stuff that had to be worked out by human engineers.

These days if you created a really advanced machine learning algorithm, I suppose it won't be long before we start seeing

robots that teach themselves how they need to walk. And we've seen some examples of that, and they look really weird, but these are mostly simulations, like we've seen computer simulations of what it would look like, where there's a virtual robot that has virtual statistics including its virtual weight and height and all these sort of things that have real world impact if you were to create an actual robot, And then there are simulations that show how a computer

works out how this robot needs to move in order to actually walk without falling over, and in some cases the animations are hysterical and also nightmare inducing. But as far as I know, we don't have any actual physical robots that follow that model. I could be wrong about that. By the way, there's advances in robotics all the time, and they're very well. Maybe a project out there that's much further along than what I'm suggesting, but the ones

I have seen of all been virtual representations rabs. Robots. By the way, we're mostly dynamic robots, meaning they were dynamic in the sense they needed to move in order to remain stable. You could have static stable robots, but these were ones that needed to move in order to maintain that upright position and not just fall over. And even some of those early robots, which often were tethered to control and power systems, could do really amazing things.

So most of the early robots, if you see the footage from Boston Dynamics, they have all these different cables and stuff coming off of them. Those cables are frequently going to power systems and to controls the dumbs. Because the robots were meant to demonstrate specific types of locomotion, they weren't going to be used as a finished product. It was more about studying the way to build the best systems, and so you didn't have to worry about

making something that was all self contained. You were studying all of this in the lab anyway. So a lot of those videos you'll see just have the this like nest of chords that are grouped up over the robot and go into some or you know, system or other multiple systems in some cases. But even in those early robots, there's some video clips of phenomenal robotic gymnastics, Like there's somewhere engineers are controlling the robots and making them do

things like forward flips while traveling a path. I'm talking about like hopping robots. They'll be hopping in a circle and then they'll do an extra tall hop and do a full forward flip and land on their foot and catch their balance and continue going in the circle. This is decades before we saw the viral clips of youtu both Big Dog and the other famous robots from Boston Dynamics.

So if you get a chance, do some digging on YouTube for Boston Dynamics and take a look at some of those early videos, because they're really really impressive and in some cases extremely creepy. Another Raybors contributions on the list dates from when he and his team developed quadruped robots that can move at different speeds like trotting, pacing, and bounding, and they could switch between those different modes so you didn't just have to turn it on and

then put it down on the ground. You could actually go from one speed to the other and you know, knock up a notch or down a notch. If you watch slow motion footage of animals moving at different speeds, you'll see that their legs move differently at each rate of travel. So like a horse at a trot and a horse at a gallop, their legs are behaving in different ways. Well, Raybor was working very hard to create

the same sort of thing in his robots. So Rayber and his team worked to push the envelope in the field of robotic low emotion. They experimented with different designs create more stable, capable robots, and many of those early designs used humans to guide the path of the robots, so it was kind of a partly human controlled, partly

robotic controlled system. Frequently, the robots themselves had computers, either on board or just off board that did all the actual work to determine the particulars like where does the leg go and how would it need to shift its weight in order to obey the commands. So the human is essentially giving commands like I want you to move over forward five feet, but it was the robots computer that was actually doing the work of how does it

achieve that goal. In n Rayber founded a company to continue this work and find commercial applications for the technologies he was developing, and that company was Boston Dynamics. So we're finally at the point where Boston Dynamics is its own company, which means it's time to take a quick break and thank our sponsor. All right, So it's the early nine nineties and bust and Dynamics as a new company. So do they start turning out commercial robots right away? Nope.

As a matter of fact, as of the recording of this podcast, none of the company's work has ever been turned into a commercial product. Directly, you cannot go out and buy a Boston Dynamics robot for yourself. But the company's work has been extremely important to push forward the evolution of robotics, and there are some really big organizations that have been extremely interested in that, such as the

Department of Defense. Now, much of Buston Dynamics funding has come from really big contracts awarded to the company by the Defense Advanced Research Projects Agency, also known as DARPA. And you might remember DARPA as being the group responsible for the Grand Challenge, in which self driving cars raced through courses ranging from desert landscapes to a mock up of a working city to see who could come out

on top. Or maybe you remember brit DARPA as the agency which at the time was known as ARPA that funded the development of technology that became the backbone of the Internet. We wouldn't have the Internet as it stands today without the R and D arm of the Department of Defense. As it turns out, DARPA has been very much interested in robotics, and the US Army, Navy, and Marine Corps have similarly been interested in potential applications of

robots that are either inspired by man or beasts. So Boston Dynamics has frequently operated as a research facility on behalf of these organizations. Now. One of the earliest projects that we ever got to hear about from Boston Dynamics was Big Dog, dynamically stable four legged robot that debuted in two thousand five. It was a predecessor of a larger project titled Legged Squad Support System or LS three.

Boston Dynamics partnered with the Jet Propulsion Laboratory, the Concord Field Station at Harvard University, and a robotics company called Foster Miller on this The concept behind the design was a robot that would be able to carry field equipment in support of deployed military squads. So rather than load down soldiers with dozens of pounds of gear, the robot would be able to carry those loads and free up the humans so they didn't have to tire so quickly

and they can move around with fewer restrictions. Dr Martin Bueller, formerly of McGill University, led the Big Dog project. Dr Bueller would be the Director of robotics Research at Boston Dynamics from two thousand three to two thousand eight, but would eventually move on to work for another corporation, one called I Robot in two thousand and eight, and then later on would move on to other projects. Big Dog could carry forty or more than ninety pounds. The robot

itself weighed in at a hundred nine ms. It was a meter tall, and it wasn't some battery operated robot. Instead, it actually used a gasoline fueled engine. The joints used hydraulic actuators, meaning the robot would use liquids under press sure to extend limbs or reduce pressure to allow limbs to contract to bend backward. For sensing its environment, Big Dog also had stereo cameras so it could perceive depth, and a light oar system to get an idea of

the orientation of various elements in an environment. Now, a light OAR stands where light detection and ranging. Here's how it works in a nutshell, a light ar detector has two main components. First is an emitter or a light gun of some sort, which shoots out light, typically in the infrared range. The light shoots out from the emitter until it hits some sort of object, and then some of that light bounces back, and that's when light ars.

Second big component, a detector picks up the returning signals. By doing some quick math, the lightar system can determine how far away the object was based on the delay between the emitted infrared light and then the detection of the infrared light when it comes back. This is the system used in modern speed guns. Oh and in a speed gun, the emitter typically shoots out a burst of

these infrared way wives. So when the emitter picks up the returning waves, it can look at the difference in distances from the emitter and thus and for the speed of the object that the gun was pointed at, such as a super sweet red sports car, because we all know the red ones go faster. This is also the robot that could keep kicking and keep on ticking. I guess anyway, there are many videos showing people giving big Dog a let's call it a healthy shove with a

foot to throw the robot off balance. You've probably seen one of these videos at least where there's this four legged robot and some dude comes up, puts his foot right up against the robot and gives us a big old shove, and the robot just stumbles around for a

bit before picking itself back up. Now, the purpose of those videos isn't a show that robots, hey, we're still the boss of you, but rather to demonstrate the ability of the robots to catch themselves even after they've been shoved, and to regain their balance through careful, though sometimes seemingly haphazard placement of their legs. Now doesn't work every time, no, because terrain can be really unpredictable. Sometimes the robots best efforts just aren't enough to keep it from toppling over.

But there are plenty of video examples of those robots, which look fairly lifelike in their locomotion getting kicked before regaining their footing. And it shouldn't come as any surprise that the robots movements look lifelike since the engineers were frequently relying upon animal movement as a model for robotic movement. As for the l S three project, it added some more requirements. This is the bigger version of Big Dog.

Sometimes they would call this Alpha Dog, and it could carry up to a hundred two of gear for a distance of thirty two kilometers or duration of twenty four hours. According to Boston Dynamics, during one control test, it was actually able to carry five hundreds of mass it's more than a thousand pounds. It also could recognize a designated leader, so an actual human being, and then follow behind that

specific person that leader without needing any other operators. On top of the sensors that it had from its a smaller cousin Big Dog, it also had GPS, so it had some additional sensors inside of it. And ultimately, though it wouldn't go very far beyond a few test missions with the military. So the advances in robotics locomotion were considerable, and there are many videos showing how impressive this robot is and how it maneuvers across different types of terrain.

But it had a big, big problem, and that was it was noisy. The robots gasoline engine created a huge amount of noise, which meant that if you were a military squad and you had one of these in your group, your position would immediately be given away. There was nothing stealthy about it. Boston Dynamics would create a smaller, battery operated version of that robot called Spot, But spots problem was not noise, it was just carrying capacity. It wasn't as hefty, It wasn't as strong as Big Dog or

Alpha Dog. It could only carry forty pounds or about eighteen kims, which wasn't really considered to be enough to be useful to the military, so it was much more quiet but had less utility. Both Big Dog and Spot would come to an end. Their projects would end, sometimes around two thousand and fifteen, but the progress made during the design and build out of the robots would inform future efforts at Boston Dynamics, So it wasn't like it

was a total loss or anything. The value was really in the actual design and build out of these robots, and that becomes a story, kind of exploratory research, kind of story for Boston Dynamics throughout its entire existence. Another robot that would come out of this effort would be spot Many. This is a much more recent robot. It's a smaller variation on that four legged design spot Many debuted in two thousand seventeen, and it has a much

more friendly looking appearance. A lot of the you know, Big Dog and Spot were kind of industrial and in looks like they weren't they weren't meant to be pretty. Spot Many has a little bit of a friendlier look with some three D printed parts, and it makes a little more sleek and it also has a nice bright yellow color to the ones that are on the videos anyway, um,

and it's also smaller. Buston Dynamics called it their quietest robot to date, and of course it operates on battery power because hydraulics would be very loud and also unnecessarily strong for such a small form factor. The robot moves in a very fluid, organic way, and certain motions kind of give off the illusion that it is actually alive, but we have been assured by Boston Dynamics that as not the case, so I'm gonna trust them. The robot weighs as the base model, but you can also get

one while you can't. But the Boston Dynamics also makes one that has an articulated arm that is mounted to the top of the robot. So think of the robot is like a platform. You've got four legs attached to it on the top front part of it. You've also got an articulated arm that can reach out and grab stuff is on a joint, so it's like it's got an elbow. And this adds an extra five kilograms to

the weight of the robot. And if you just take a look at the design, the hand and arm makes it look like it's kind of a head and neck of an animal, so kind of like the spot Mini is actually a tiny frisky a pedasaurus or something. According to the company's spot Mini can prance around for about ninety minutes before it needs a recharge, and it can carry about fourteen kilograms of payload and has seventeen joints of articulation. If you haven't seen videos of that thing

in motion, I highly recommend you check it out. They are fascinating and again a little unnerving because you know, they again are mocking organic creatures. So there's this sort of kind of Uncanny Valley thing going on with these robots that move like living things, but clearly are not living things. There's also Little Dog, which is a much smaller robot that Boston Dynamics created for DARPA. DARPA, in turn, relies on Little Dog as a development platform for other

advances in the robotics. So if you've listened to my other episodes that involved DARPA, you know that DARPA itself does not conduct original research. Rather, it's an agency that offers up funding to other companies and organizations to do that kind of work, and that's in the interest of the Department of Defense. Little Dogs Men is kind of a standard platform upon which robotics organizations can develop new

technology for DARPA projects. So that might involve things like pathfinding, object recognition, uh, you know that sort of stuff, things that are maybe more on the software side of things than the hardware side of things. Other four legged robots built by Boston Dynamics include the Cheetah, which could hit a top speed of twenty eight miles per hour that's about forty five kilometers per hour. The Cheetah was not built to run free across the landscape. It was running

on a treadmill. It actually had tethered power supply. Eye tethered controls, so again, this wasn't a a all inclusive robot. The purpose of the Cheetah was for robotics experts to experiment with ways to build a fast moving robot. So again, this was for them to kind of build out the technologies that would then be incorporated into future designs, rather than something that was its own self contained robot. So again it kind of a development platform. Uh. It was

really fascinating to to watch though. It would run on these treadmills at blistering speeds. You'd see this treadmill start to ramp up, and at twenty eight miles per hour, it meant that it was moving faster even than Hussain Bolt. However, it couldn't balance itself, so unlike Hussain Bolt, you would not be able to Like, it couldn't take a turn to save its life. So if somehow one of these Cheetah robots were to get after you, all you need

to do is take a left or right you'd be fine. Uh. Also, it be tethered to a computer and probably couldn't move very far, so you're probably fine either way. But that's just one They did, however, use that design to build out another four legged fast robot called the Wildcat. This one was a self contained robot. Buston Dynamics showed off the Wildcat robot in two thousand thirteen with videos of

this four legged robot bounding and galloping outside. It was not tethered so it can run around and had a gasoline engine very similar to Big Dog. Pretty loud, but also pretty fast. It could get a top speed of about sixteen miles per hour, so not quite as fast as the Cheetah. If you are Hussain Bolt, you could outrun this particular robot. If you are me, you would be robot meat. I don't go faster than sixteen miles per unless I'm either inside another vehicle or falling off

a cliff. There is another Cheetah robot by the way, if you do searches for Cheetah robot, there's a different one that comes out of the m I T Robotics apartment. This one's a really cool robot. It's a four legged robot that can detect and jump over obstacles. There's some great videos of engineers showing it uh seeing an oncoming obstacle on a treadmill and leaping over it. Even as the obstacles grow taller and taller, it's still able to get over them. It's an incredibly impressive robot, but it

is not a Boston Dynamics robot. Then there's Rise, which is a wall climbing robot. This one has six legs and it's pretty small. It's about nine point six inches or long, has a tail like protrusion that makes it look sort of like a robotic horseshoe crab, and each leg has two joints, which means it has a total of twelve actuated degrees of freedom. The feet have a little micro claws on it, so it can grip to surfaces.

As long as the surface isn't completely smooth, it can grip onto it and allow this little robot to climb right up a vertical surface. On top of the normal sensors that you would expect in any given robot, there were some additional ones to ensure that it wouldn't take an unplanned tumble, which included foot contact sensors so the robot would actually know when its claws had been more

or less engaged in whatever surface it was climbing. They also had legs strain sensors so if any legs were having more strain on them than normal, it could either redistribute its weight in an attempt to continue its climb, or it could stop its climb rather than risking falling off of its surface and allow someone to come up

and retrieve it. According to Boston Dynamics UH, this robot could actually climb faster than an average human, and unlike humans, it doesn't tire out, although it could run out a battery charge, so that's sort of similar. And this was not built for a specific purpose. It was more like building out capabilities of robots with the ultimate goal of creating robots that could be useful for people like first

responders in various UH emergencies. So there wasn't a specific mission, but it was more about let's build out these capabilities that could come in handy in future robotic designs. I got a lot more to say about Boston Dynamics, but before I get into this final segment, let's take another quick break to thank our sponsor. Okay, so let's talk about a couple more robots before we get into some

interesting recent stories involving Boston Dynamics. There was also a robot called Rex and that's spelled r h e X. That's a six legged robot designed to be able to tackle particularly tough terrain. It's not a big robot. It is about fourteen centimeters tall, so it doesn't exactly tower above the landscape, but it's curved legs could propel it over lots of different surfaces better than robots much larger

than itself. It relied on TELA operations, which means you know, someone's controlling it remotely using a camera to get a robot's eye view of what was in front of the critter. The legs rotate kind of like wheels, so they're these

sort of semi circular legs. They don't make a full circle, but they do rotate around and they can go through pretty much anything like that shows robots going through mud and water, and because it's a fully sealed robot, it doesn't you know, it's not in danger of having water, give it any short circuits or anything like that, no disassemble. It can go over rocks and twigs and even logs and stuff. So it's a pretty interesting and robust looking

little critter. And again it's sort of a proof of concept of technologies that could be used in future robots. There's also sand Flea, which is a wheeled robot, an unusual thing for Boston Dynamic or Boston Dynamics, I should say. It can jump actually, so it's a wheeled robot, but it can also leap into the air to the tune

of about thirty feet in the air. It uses gyroscope hopes to maintain its orientation during jumps so it doesn't just tumble to the ground and you know, crash into lots of little bit so it can actually land on its wheels properly and then jump again. There's some great footage of it jumping from the ground onto the top of a building that I was really impressed by, so you should check that out to sand Flea is the name of that one. Then we get to the humanoid

robots that Boston Dynamics has built. Now, the ones I've mentioned earlier were mostly based off of animal designs, with that exception of sand Flea because that's a wheeled robot. But these are all bipedal machines that I'm going to talk about at this point. First, there's the Protection Ensemble Test mannequin, also known as pet Man. This is an anthropomorphic robot that's pretty basic. Its movements were largely influenced by the work done on the Big Dog project. In

other words, it lumbers around a little bit. It when it's not wearing anything, it does not look terribly human. I mean, it's basic human shape that's got legs and arms. But it does important work for humans, and that is to test out chemical protection suits to make sure they are of suitable durability for use by real, live human beings.

So a suit might protect against chemicals when it's nice and new and when it's just kind of folded up, But on the real world, a suit's gonna be exposed to stresses and wear and tear, and so pet Man could simulate that by moving around while a suit is actually on the device. It would be wearing one of these chemical protection suits, and it could test the durability of the suit and actually even do this while being exposed to different types of stresses like temperature changes or

even chemicals. You could spray chemicals on this thing, because it's a robot, you know, you wouldn't want to do that to a real human. You'd put them to way too much risk. If in fact, the chemical resistance suit didn't live up to its design, but with a robot, you don't have to worry about that, So that's exactly what they would use it for. It's kind of kind

of interesting. Um and it's also fun to watch videos of it wearing the suits because then it just looks like a It kind of looks like a slasher movie monster, like Jason or Mike Myers, Mike Wires. That's just this big, lumbering hulk of a thing and uh typically camouflage chemical suits. It's kind of terrifying. Then there's Atlas, the agile anthropomorphic robot. The first generation of Atlas was a pretty monstrous beast itself.

It measured six ft tall uh three centimeters in other words, and waited a hundred fifty ms or three thirty pounds. It's sensor's range finders and optical cameras communicated with a computer that was separate from the robot itself in the original prototype model, so in other words, all the computational stuff was offloaded onto a nearby unit. According to Boston Dynamics, the robot had a total of twenty eight degrees of freedom when you added up all of its various actuated joints.

The earliest version required a tethered power supply, and then later builds would incorporate the power onto the robot itself, and its movement was largely based off the results of developing pet Man. So while pet Man was kind of an evolution of the work that was done on Big Dog, Atlas was an evolution of the work that was done on pet Man. And Atlas could do some pretty cool stuff.

It could balance on one leg, which is really remarkable when you think about how big this thing was, and also it could recover from being shoved uh or sometimes it could. Other times it would lose its balance and unpredictably fall over, sometimes without even being shoved. First, it was that generation of Atlas that would be featured in the Team DARPA Robotics Challenge. I did a full episode about the Robotics Challenge, but I would give a quick summary. Here.

You had a bunch of teams that were competing in designing a robotics solution to a series of problems that simulated the sorts of task a robot might have to complete in the wake of a disaster. The inspiration for the challenge was a real world disaster The Fukushima nuclear reactors when they malfunctioned after its tsunami um It was very dangerous to send people into that area, but a

robot would be a much better choice. So this was kind of a challenge to see if any teams could design robotics platforms that could respond to the sort of things that would be necessary in another disaster of that caliber. So the robots would have to operate a vehicle, a human powered vehicle, typically not a human power, but human operated vehicle, to a specific destination, so in other words,

like a golf cart. They'd have to drive this themselves, get out of that vehicle, open a door, walk through a doorway, and then complete several other challenges, including the operation of a handheld power tool. At one point, six of the competing teams would use the Atlas robot as their robotics platform. Other teams would rely on different humanoid robots, so they weren't required to use Atlas, but six teams

did use Atlas. The nature of the challenges meant that the robot would have to simulate human activities, and the winning team, Team KIST, used a robot called Hubo, so it was not an Atlas. There are a lot of videos of the competition out there, including some of the various Atlas robots. There are also some great videos of heartbreaking and occasionally incredibly funny moments of robots falling over while trying to do seemingly simple things like walk through

a doorway. As it turns out, that's a lot trickier than it sounds. In two thousand and sixteen, Boston Dynamics unveiled a second generation of Atlas robots that looked slightly less industrial and had more advanced sensors for navigation and orientation purposes, as well as more advanced maneuverability technology. It could walk on surfaces like snow, and even planned paths in areas where other objects are currently in motion. This one was a seventy five tall, so a little shorter.

That's about five ft nine inches tall. It two ms or around a d eighty pounds, a little you know, significantly lighter than three pounds. Boston Dynamics has credited three D printing for the drastic reduction in weight and size, and there's an even smaller, lighter, and more capable generation of Atlas the debuted in late Boston Dynamics posted a

video of it showing this robot in action. It could jump up on boxes, so it could walk up to an obstacle like a box, judge how tall it was, and then jump so that it landed and maintained its balance on top of the box. It could even do a full back flip off a box and land on the ground, regaining its balance so it doesn't even tip over, which is an incredible thing to see. If you haven't

watched that video, you should definitely find that one. And that brings us to handle a robot that seems to look like kind of like a bucking horse on roller skates. I've seen other people refer to it as kind of like a donkey on roller skates. It's a bipedal robot, but it's lower legs don't end in feet, they end in wheels, so it has two wheels at its base, kind of like a segue, and it uses that same

sort of technology to maintain its upright position. Has two upper limbs that can pick up and carry objects weighing up to and the robot is two meters tall, weighs a hundred and five kilograms, and has ten actuated joints. It uses battery power and a combination of hydraulic and electric actuators to actually work those limbs, and has depth cameras to provide perception. It can speed along the ground

and even jump up onto higher surfaces. One demonstration, there's a video that you can watch of this as well. It will show the robot rolling toward a table and then the last second it jumps up and lands on the table surface, rolls across the table before landing on

the other side and continuing on. It's pretty impressive, and Boston Dynamics has also pointed out that because it only has ten actuated joints, it's far less complex than some other models, and that uh that reduction in complexity could mean a reduction in cost as well. The wheels work on most services until you get into some really difficult terrain, and I find it fascinating that a company that had made its name from mostly legged robot designs had gone

with this hybrid approach. But the videos show that the design is pretty effective, at least for specific applications. Now, one thing I have not really covered in this episode is who owns Boston Dynamics these days, because that's actually changed a couple of times over the past few years.

It operated as an independent company for more than a couple of well a couple of decades, and then on December thirteen, the top secret R and D branch of Google, known as Google X and these days just called X acquired Boston Dynamics um as X it is now a subsidiary company of Google's parent company, alphabet So. At the time, it was a division under Google. Now it's its own

spin off company called x SO. Boston Dynamics was just one of nine robotics companies that Google acquired around that time, and the co founder of Android, Andy Reuben, was slated to take the helm of all nine companies under a new robotics division that was called Replicant, which isn't terrifying

at all thanks a lot Blade Runner well. According to Reuben, the initial plan was to have those companies do their own thing, working on research the way they normally would and advancing the art and science of robotics, so in other words, there wouldn't be any pressure to develop commercial products right out of the gate. Instead, they would just do exploratory research and the advances in the field would guide Google's decisions on how to leverage those developments in

commercial applications. But in October, Andy Reuben would leave Google. Now Reuben reportedly left to found a new company to act as an incubator for technology hardware products. So, in other words, he wanted to build a company that would help hardware developers get to market with their ideas. That

company is called Essential. But later in information would leak that Reuben had apparently or allegedly i should say, left Google after a one of his employees, a woman, had filed a complaint to human resources about Reuben having an inappropriate relationship in the workplace. Now, Reuben's spokesperson over it Essential says that this report is misleading, that Reuben had never engaged in any misconduct at Google, that any relationships he had at Google were consensual, uh, and that there

were no other issues there. But whatever the case, Reuben took a leave of absence at Essential to deal with quote personal matters in the quote. So back to the robotics companies under X. So Reuben leaves the company, and there was kind of a vacuuman leadership at that part of Google, and the is a lot of confusion about what was going to happen after all these companies had been told that they were going to continue operating as normal,

really doing this exploratory research. But in Google management attempted to put some sort of leadership and structure in place, and reportedly the change came with an intent to make actual commercial products, so the divisions would have to dedicate themselves to building stuff that could be sold. The mission statement was effectively being changed from do exploratory research to

build us something that we can market now. According to Business Insider, Boston Dynamics was told to start working on designing consumer robots. Now. These would have to be quiet, and they'd have to be suitable for operating around human beings of all ages. And this was a really narrow focus and something that many people at Boston Dynamics were

not necessarily keen to work on. For one thing, they wanted to continue their work in pushing the boundaries and looking into all sorts of mechanisms to operate robots, including ones that would not be suitable for a consumer robot. For example, hydraulic actuators. You know, they're much louder than electronic ones, but more powerful, and therefore they can give a lot more uh uses applications for robots, but ones

that wouldn't necessarily be appropriate for a consumer product. Jonathan Rosenberg, who had taken the helm of the replicant division, stated, quote, we, as a startup of our size, cannot spend thirty plus percent of our resources on things that take ten years end quote. So he was saying, we can't really afford the luxury of doing exploratory research with the idea that sometime down the line it's going to pay off. We have to start making things that we can actually sell.

So from a business perspective, you can kind of see where he's coming from. But from the perspective of people at Boston Dynamics, this was very antithetical to what they had been doing up to that point. So Boston Dynamics would eventually go on to release a video of Atlas stacking boxes in an application that might be suitable for something like a warehouse, But apparently that quote soured the

soup end quote with Google. Google management wanted a consumer facing product, not something that was going to have applications in industrial settings, and so at that point Google began to shop around for another suitor to take over Boston Dynamics. Reportedly, one of those potential parties was the Toyota Research Institute, but ultimately Google found a buyer in a Japanese company

called soft Bank. Soft Bank also bought another robotics company off of Google slash alphabet called Shaft, which had competed in that DARPA robotics challenge I mentioned earlier. Masa Yoshi, san CEO of soft Bank, released this statement when news broke that the company was acquiring Boston Dynamics. Quote, today, there are many issues we still cannot solve by ourselves

with human capabilities. Smart robotics are going to be a key driver of the next stage of information revolution, and Mark and his team at Boston Dynamics are the clear technology technology leaders in advanced dynamic robots. I am thrilled to welcome them to the soft Bank family and look forward to supporting them as they continue to advance the field of robotics and explore applications that can help make life easier, safer, and more fulfilling. So that's the story

as it stands now with Boston Dynamics. There's a lot of hope over at Boston Dynamics that they will be able to continue their mission that they had been following for years before the Google acquisition. There's also obviously some concern that maybe soft Bank will follow suit with Google and say, hey, can you make us a robot that you know, we'll bake cheesecake and greet people at the door. And I totally want a cheesecake robot now. And I need to think these things through before I say them,

because this was not in my notes. It was just kind of off the cuff. And Okay, guys, we're gonna wrap this up. So that's the story of Boston Dynamics so far. I can't wait to see what comes out of that story next. But if you guys have suggestions for things I should cover in future episodes of tech Stuff, you should write me the email addresses tech Stuff at how stuff works dot com, or you can drop me

a line on Facebook or Twitter. The handle of both of those is tech Stuff hs W. Remember we also have an Instagram account, and you can watch me record shows live on Wednesdays and Fridays over at twitch dot tv slash tech Stuff. There's a schedule right there. You can check that out and chat with me in the chat room if you like. I'll enjoy that interaction. I'm wrapping this up because there's a cheesecake out there with

my name on it. I gotta go find it. And you guys, well, I'll talk to you again really soon for more on this and thousands of other topics because it has staff works dot com zo