Raptor robot runs at 28.58 mph, faster than any human [video] | Hacker News Reader

Raptor robot runs at 28.58 mph, faster than any human [video]

kurzweilai.net

81 pointsca98am7958 comments

These are really impressive from a mechanical engineering point of view, but in all these demos the powerplant is not on the frame. Instead, power is fed via wire, making it stationary. Its a little disingenuous to claim it goes 28 mph when it doesn't even power itself. I'd love to see what real speeds they can get out of these things with a powerplant installed, especially over mixed terrain.

Right now, this is like running a Tesla with a very long extension cable and claiming absurd 0-60 times. Instead, tell me how well it operates with a battery or an engine and all the extra weight and engineering accouterments that requires.

> These are really impressive from a mechanical engineering point of view

It's not just a mechanical engineering problem, the dynamic of the system is way too complex to be modeled with a simple formula. There's a lot of AI/machine learning involved in robotics. That's hard with or without external power supply.

I'm going to reply to your comment, as it's currently on top, but several of the top comments have this same sentiment.

I don't think it's fair to criticize this project harshly because the robot has external power and a supporting arm to keep it from damaging itself. The research and development was clearly about the design and actuation of the legs, and the researches have clearly done well on that front. To require them to provide onboard power for a system that they don't even know the requirements for is a little much, in my opinion.

The research and development was clearly about the design and actuation of the legs...

OK, but then is it really fair for them to publicize how fast it can run? I mean, I guess if the challenge is just to be able to move the legs fast enough, then maybe it is. But then why not publicize the number of steps it can take per unit of time instead of its speed? Speed is kind of misleading anyway unless you know how big it is.

Yes, it is fair to disclose exactly what you've achieved. They've made it run 28.5 mph on a treadmill with off board power and that is exactly what they disclosed. I guess you are reacting to the comparison to a human that has on board power. Or maybe it's just the fact that this comparison makes for a misleading headline. In any case this is probably the least interesting thing we could be discussing about their progress.

Clearly there are two parts to this equation.

One aspect is how far research has gone - it's gone a long ways and achieved a lot, yes.

The other aspect is how there is to go - there's a long to go before you get anything useful here.

The thing you have to consider with these criticisms is that many technologies have gone a long, long way in their research but still hit a wall before they wound-up being practical. Walking robots seem like they're just very hard. Compared to wheeled vehicules, "walkers" touch on a very large number of "frames" (in the sense of Minsky etc). There's no reason for a walker if they just use to the safe, level zone of a road and if a walker moves in less structured areas, they need a lot of intelligence to move in a way that's useful for people (I've seen a video of BigDog moving through the woods with one soldier guiding via joy but the useful of thing that needs constant attention but can't be ridden would be strictly less than that of a horse and that's being optimistic it can guided effectively, etc).

I suppose once you have the basic mechanism working at a few different scales it's not such a big deal to hang the battery off the center of gravity. It's true that this project was about finding good 'leg' designs and balancing mechanisms; at the same time I'll be more impressed when they present a free-standing version that be taken to an athletic track.

I think it won't be long (<5y) before someone does this with Usain Bolt or whoever else is the fastest human at the time. Shades of John Henry.

   development was clearly about the design and actuation of the legs, and the researches have clearly

..failed by investing time in blind alley. This mechanism is unable to:

-move legs out of sequence

-support turning or balancing in Z axis

its like one of those "running" robots that pretend they arent wheels

https://www.kickstarter.com/projects/138364285/outrunner-the...

FWIW, the snark in my comment wasn't directed at the research but at the reporting ("faster than any human!").

Eh. That's kind of like complaining that a human cheats by eating plants and animals. If you really wanted to show an example of something walking, it should have a nuclear power source, or have photo cells to absorb sunlight...

But yes, I get your point. It'd be nice if they ALSO showed it carrying its own power source. I believe that would just lead us to questions of "how far can it go with a given power source, at a given rate?"

Not so much you can have vary light weight power plants if you sacrifice distance traveled. Going X MPH over 100 feet is less interesting but would not change any of the hard parts.

More importantly an ostrich can do 45 MPH which is far faster than this.

> More importantly an ostrich can do 45 MPH which is far faster than this.

It's also much larger. I imagine once they scale this up, it'll go even faster.

Can you imagine how frightening it would be to be chased by one of these things?

I, for one, welcome our new robot-raptor overlords.

Imagine it car sized, and running on a serial-hybrid power plant (electrical for sprints, turbo diesel that spins up while it "rests" to recharge its battery for sprints).

Terrifying.

I'd be terrified even by a little one(consistent with the current size). If it had teeth, or that spinning balance beam had blades on it and they get maneuverability down...boy. Here's hoping DARPA doesn't fund the project to turn it into a weapon.

(edit: just to be clear - i can't run anywhere near as fast as usain bolt, so it wouldn't have to be ostrich or car sized to catch me)

They aren't studying powerplants.

The complete package is important for a finished product, but this isn't one.

I wonder how the treadmill affects its stability. It doesn't need to overcome its own momentum in order to accelerate; rather, slowing down requires acceleration.

That's a great leg imitation and obstacle tolerance but does it really count as running when it's permanently attached to a stationary object?

I was thinking that too. I bet Usain Bolt could run faster if he had an arm holding him up.

A lot of people seem to take issue with the lag of on-board power. OutRunner i a running robot with on board power, that have reached some impressive numbers while running outdoors.

The projekt is on kickstarter right now: https://www.kickstarter.com/projects/138364285/outrunner-the...

its a WHEEL, stop with the running nonsense

http://11even.net/wp-content/uploads/Mutant-bicycles-1.jpg

I think people are missing the point about the attached boom arm. They're working on the mechanics of the bipedal motion before worrying about making an autonomous robot.

Randall Munroe's nightmare has come true.

Have you ever created something which requires somewhat esoteric knowledge to create and understand, and then you show it to someone who doesn't have that knowledge, and you find yourself disappointed with their reaction, and disappointed that they don't understand enough to know that they don't understand? Made up example:

"Hey look, I designed a CPU and created a basic computer and wrote an OS...it's basic I know but..."

"Can it run Call of Duty? Xbox is better"

"That's not the...point...nm"

That's what this thread feels like. Except I would have expected HN commenters to have enough insight to know that they don't understand enough to be commenting.

Link to video. If there was embedded video on that page, it wasn't working for me. Here is a video I found of it on YouTube:

http://www.youtube.com/embed/lPEg83vF_Tw

I'm a little disappointed that its tethered... Then again, having little robots that run faster than any human be tethered is probably a good precaution just in case we ever achieve singularity.... ;)

Yeah, but Usain Bolt has on-board power.

No on-board oxidizer storage. Less than a weeks worth of water storage. Lame.

Are we supposed to ignore the arm the "robot" is attached to? Set that thing out on a road and let it run and I will be impressed.

Not quite two legs, but you might like this :) http://www.dailymail.co.uk/sciencetech/article-2626036/Move-...

So does a car. Whats your point?

I think the fact that cars don't use bipedal locomotion to achieve fast velocities is the point.

I am not aware of any bipedal cars, and I think that's the point here. A machine that moves like human, yet for the first time ever can actually do it faster than we can.

It doesn't move like a human. It has digitigrade legs and a forward-backwards horizontal balance thing. It runs, I guess, like a raptor.

It's not a big shock that things built for speed can run faster than humans can. Horses can also run faster than humans. So can cheetahs. Ostriches, if you want something bipedal.

Exactly. It also can't move in any of the other ways a human can, it certainly won't be climbing obstacles.

It's not as if humans are the pinnacle of power efficiency. We're versatile, but inefficient. So if you're going to compare a machine to human movement, it should be on the basis of versatility. A car is way more efficient than a human at moving on a flat track.

A human uses less energy going a mile on foot than a car does. So for one definition of efficient, the human wins. They should still win even if they are carrying a modest load.

I guess I would have used effective where you said efficient.

(If you add in the energy used by modern agriculture to produce the food the human ate, the car probably wins: http://ideas.4brad.com/holy-cow-walking-consumes-more-gasoli...)

How much energy does a human use when he carries a car for a mile?

I'm curious how I could have better phrased my comment so as to not invite responses like this. I guess I realize that 'efficient' vs 'effective' is a quibble, but I was trying to respond to the ambiguous usage of 'efficiency', not trying to give the final answer on the comparison.

Sorry, I realized you were trying to be circumspect, I just was instantly visited with a vision of a guy trudging around a track with a car on his back.

How much of that comes from bipedal motion, vs the way that humans break down their food?

http://www.youtube.com/watch?v=aIwCwwrDnGI The wheel of cheese rolls effortlessly down the hill with no power source, while the bipeds all end up turning into makeshift wheels trying to catch up to it.

Simple really, robots will do most physical jobs better than us and society needs to adapt to that. They will also be used in tasks we find disgusting like military conflict. Seeing how little real outrage drones deaths are causing; mostly because of deft political handling of the press; do you think employing robots against "other people" really is going to be much different?

The best outcome is that the onset of robots doing much of the work is that it frees enough of the world from economic strife to end the need for warfare.

Biologicals still have an edge - the power-plant issue. Until batteries get a LOT better, or fuel cells get cheaper, a person that lives off of peanut butter and lemonade is a real advantage.

Especially since a very active human will only require the energy equivalent of a liter or two of gas per day.

> frees enough of the world from economic strife to end the need for warfare.

While this sounds good, I'm not sure this would happen for two reasons:

- we'd have to completely change the system of work / salaries / resource allocation - the way we're going right now we'll end up with smaller number of companies providing services / resources to huge society that is much larger than needed to support those companies

- I don't believe people have a need for warefare... they may actually be more interested in it, assuming their other daily needs are satisfied in an easier way

Why do you think automating much more work would make life better for people in general?

The world has economic strife not in the least because of greed, and violence not in the least because of the lust for power. I am afraid more efficiency might not necessarily reduce those, it might even enhance them.

It's a pity this was downvoted because that's exactly the sort of thing against which new tech should be tested.

So, to really answer the 'what's your point' question I think the point is that bipedal could be more energy efficient than wheels and that feet can go where wheels can not.

> I think the point is that bipedal could be more energy efficient than wheels and that feet can go where wheels can not.

The second one, absolutely. As for the former, I don't think you can get any more efficient than a wheel on a flat track - certainly biped movement is a lot less efficient since you need to work against gravity.

The perceived efficiency of bipedal movement comes mostly from the ludicrous efficiency of biological actors when converting chemical into kinetic energy – a cyclist (or even runner) can go a day on the chemical energy that would get a motorcycle to the next petrol station [0].

Also consider that even the fastest bipedal runners tend to go much slower than any motorised vehicle, and even quadruped animals rarely reach into what is considered standard with cars.

[0] The guideline daily amounts for a human are roughly 10 MJ, from personal experience, cycling a bit during the day only increases that marginally (e.g. 10-20%). It seems reasonable to claim 1 MJ per 100 km cycled, whereas one kg of gasoline (~1L) is at ~42 MJ. Given the usual 1-10L/100 km, 1 MJ appear to correspond to a few kilometres.

Here is a trivial proof of the former: take an elite biped marathon runner on foot vs an older marginally in shape biped like myself on a bicycle, and on my bicycle I can beat the elite runner quite easily.

Another way to model it is I can quite easily maintain 10 MPH without much if any sweating on a bike trail, but on a calm hike with the kids on the same trail I struggle to achieve 3 MPH average.

One key difference is friction. With wheels there is continuous contact with the substrate, which means continuous friction and thus energy loss. Since this robot has an aerial phase, this should be considered an advancement towards the energy efficiency of most land animals who only* lose energy due to friction when in contact with a substrate.

Edit: *not factoring wind resistance, which is present regardless of mode of locomotion.

Cars aren't bipedal so no, a car doesn't "run" faster than a human.

Cars can't jump over obstacles, climb stairs, or traverse rubble.

Hrm, seems like an opportunity :)

Paving, for the win!

I was originally thinking along the lines of hillclimbing jeeps. But imagine a treaded vehicle that could automatically level and pave a road. There are these things https://www.youtube.com/watch?v=jkVBg_-OviI and these things https://www.youtube.com/watch?v=3ZMPoHahaj8 but they all look like they're running on ground that has already been cleared and leveled. It would be really interesting to see something that could dig through ground and rock like a tunnel digging machine and compresses the material into tile blocks to use as pavers.

Is the "dynamic balance control" really a large gyroscope, with rotational acceleration to control falling forward and backward?

Nonetheless, I was really hoping to see what happens whet it falls down while running at high speed on the treadmill.

28.58 mph == 45.995 km/h [0], BTW.

[0] http://www.wolframalpha.com/input/?i=28.58+mph+in+kilometers...

It is surprising that such things have not been created before. It's a few legs supported on a sideways axis. One would expect it would not have been hard to do it at least 50 years ago.

But in robotics everything is about 1000x harder than one would naively assume, it seems. (Judging by how my vacuum robot with a friggin' laser scanner can't find its docking station from 1 meter away.)

Reply on news.ycombinator.com