Hacker News new | comments | show | ask | jobs | submit login
Introducing Handle – Boston Dynamics [video] (youtube.com)
429 points by pauselife on Feb 27, 2017 | hide | past | web | favorite | 119 comments



I'd like to see it fall and recover. These videos just aren't what they used to be since they stopped kicking them robots.



Handle looks like it can kick back.


Actually it looks to me as though the legs cannot move sideways at all, only forward/back/up/down. If that's true it would be nearly impossible for Handle to stand on one leg or even raise one leg for long enough to perform a decent kick without falling over.


In the video where Handle is moving in a circle quickly while crouching you can clearly see one leg moving out to the side.


Nope. The legs are still pointing in exactly the same direction relative to the body, the only difference is the body is tilted. Relative to the body the legs have different vertical extension but the horizontal distance between them is the same.


This must be fixed.


Or lasers...


IDK, seemed pretty capable when it hit that ice/snow patch coming down the hill.


I can't wait for weapons to be attached to this thing so it can be used against us in the future. Imagine the part where it was spinning, but instead with a firing automatic weapon. What prevents democracy from collapsing when a certain technical elite control the most powerful autonomous weapons and keep tabs on everyone's daily lives?


Probably a few years before they finish developing the flying machines that can be anywhere on the planet within 20 minutes to drop a bomb and obliterate the entire building you're in. Those jet engines they're talking about would be hot enough to melt tungsten, this is just math, it's trivial in comparison.

Edit: and don't get me started on the "cruise missiles". 1000 km range and they hit within five meters, and with unit cost less than training for a single infantryman? Total bullshit, I can't believe anyone buys it.


What are you saying is bullshit? I'm pretty confused.


If you want to complain about robot weapons giving people the ability to kill without personal involvement or compunction, you should have complained:

* Ten years ago, when we started arming drone aircraft with anti-tank missiles

* TWenty years ago, when cruise missiles gained active guidance

* Thirty years ago, when we got laser-guided bomb technology working

* Eighty years ago, when radio let commanders give orders to subordinates without seeing their suffering

* A hundred years ago, when artillery gave us the ability to kill tens of thousands of people without them ever leaving their trenches

* Three hundred years ago, when firearms became truly viable and infantrymen could kill their opponent from far enough away that they didn't get blood on them

Don't let the human shape make you afraid. The Terminators had human faces because that made them scary, not because it was effective. People are afraid of these robots because they have two arms and two legs, not because of what they can do, and these fears about "spinning in circles shooting machine guns" are pathetic rationalizations of their loss of reason in the face of emotion.


Someone still has to man the lengthy manufacturing and supply chains to build these things and their consumables plus someone needs to be deployed into the theater of operations to perform logistical support and maintenance to keep your hypothetical killbots in action. (Automated manufacturing is one thing, automated diagnosis and repair is quite another.)

Someone also has to have acquired enough military expertise to command these units effectively, something "a certain technical elite" feels is beneath them.


> something "a certain technical elite" feels is beneath them.

A very common sentiment. However, please remember that the skills to get in at the bottom of each field are radically different and don't let that colour your perception of the top of each field.


Additionally - If the elites make it lucrative enough it will be easy enough to get labor necessary to complete the tasks. Tribalism is real and if they were able to convince a subset of people to complete their ends then controlling the minority/subset would be easy. It boils down to basically why Nazi's and Aryan pride were successful.


yes but in the end they lost, nazi's were defeated, the freedom-loving humans won :o


The elites mentioned would just use ML to war game and hone tactics, and no one has experience fighting with (or defending against) this type of mechanized warfare.


Plenty of videogames have trained people for that. The lower end of the technical elite.


>Plenty of videogames have trained people for that. The lower end of the technical elite.

This is a prime example of the disdain for the military I was talking about. Actual warfighting is far more complex than noob tubing and zerg rushes, I assure you.


Wait until robot personhood makes disabling one of these things equivalent to murder, then we're in real trouble.


Citizens United (corporate personhood) already has laid the groundwork for that. If a corporation has the same rights as a person and an ai is a representative of said corporation, the ai would be protected under the corporate personhood umbrella.


How many muscles did you pull with that stretch?


Chances are, you don't live in a democracy anyway. Most "western" world is either a representative republic or the same mixed with a monarchy.

Switzerland is the only country that gets close to a democracy - they let their citizens vote on many issues.


A representative democracy is still a democracy.


That's technically true (the best kind of true) -- a representative democracy is a republic --, but the problem that the poster was pointing at was that the people are not being actually represented by our elected representatives.


Is it? When was the last time anybody asked you whether your country should go to war? Or how to spend your taxes?

Finding a representative that you agree with on most (let alone all) issues is pretty much impossible.


The Economist Intelligence Unit, amongst others, ranks countries on this dimension: http://www.economist.com/blogs/graphicdetail/2017/01/daily-c...


People itself are the cheapest and most versatile "killbots".


Guess we should be building our own ... to respond in kind.


yes, things are finally getting interesting around here.


> used against us

Is that us the US by any chance? Because if these things ever become war machines I doubt the US will be second to anybody in using them.


You mean like the kind that comment here and have all sorts of interesting ideas that 90% of people don't care about or would resent? I wonder about that a lot too. Like what are googlers really up to with all that control of knowledge... </scaremongering>


If you thought I was referring to Google, or any specific corporation, government or entity, I wasn't. Technical elite was ambiguous intentionally, because it really can be anyone who has the ability to fund this type of technology, forward thinking enough to realize the power of it and isn't concerned about using it against others. Historically, this doesn't seem far fetched.


Given that this is Boston Dynamics were talking about I would say this isn't just far fetched, it's probably dead on.


Has boston dynamics put these things to some real usage? They 've been showing impressive videos for 7, 8 years?


Unfortunately there is no practical application for them. They're incredibly capable, but they're also expensive, fragile, and difficult to maintain. This is, in a nutshell, why Google decided not to hold on to the company - no consumer market whatsoever.


I spent five years in software for Amazon's fulfillment centers. Let me assure you, there is a market for this sort of robotics.

I can think of a few jobs currently handled by humans in the FCs where this robot in particular could do the entire job. Lots of kinks to work out I'm sure but that's what the software team is there for.

Google would probably never sell to their biggest competitor at a reasonable price though. Plus, Amazon already has a robotics company in Boston (Kiva).


They were never actually sold, the only news related to it was back in march 2016 where alphabet said they were looking to sell. Afaik nothing has come up about it since.


and their about-us page still says that they are a "wholly owned subsidiary of Google, Inc."

http://www.bostondynamics.com/bd_about.html


That should be Alphabet by now.


that is not why google decided to "not hold on to the company". it was far more due to culture and direction conflicts. in general, it seems google does not really understand what to do with the companies it purchases.

http://www.businessinsider.com/why-google-and-boston-dynamic...


For a while they were under military contract for these, but the versions they built were expensive to maintain, but also much too loud. The military contract was removed, but i think as a result they worked towards making them more simple and more quiet.

In the end though they're still building prototypes as R&D to try and find something that can be polished into a product.

here's a video with commentary: https://www.youtube.com/watch?v=giS41utjlbU


Presumably they might be learning some really cool stuff building these prototypes that could be used for something more commercial, or licensed.


I wonder how many cars in the parking lot have been trashed at this place while working out quirks with these robots. I think I would be parking down the road a bit.

ETA: I guess that's skin in the game. They probably have an agreement that everyone involved with the project parks their car in the parking lot for the first parking lot run.


A few years ago they were testing one of the big ones they built for the military called the legged squad support system. They were testing it in the parking lot, and it ended up rolling over and totaling this guys car. Turned out it was the guys first day there.


> I guess that's skin in the game

Should be a requirement in most engineering/maintenance fields IMO. When I used to be a pilot, the engineer who signed off on the maintenance manifest after a major aircraft service was always required to go up on the check flight with the flight crew.


That 80's jump-freeze-frame was fantastic

I want to see them take this to a skate park


Someone on reddit mentioned it's probably a homage to the end scene of Short Circuit


Same thought! I actually thought it would start grinding a rail after the first jump.

Next up: Tony Hawks Pro Skater Robot edition


If I'd seen this in a horror movie 10 years ago, I'd probably have laughed. There is something about it being real that truly terrifies me.


I felt the same way. I don't know if it's the fluidity of its movements while it was moving or something else. I kept thinking of Silent Hill for some reason.


The warehouse setting, and then snowy outdoors. Just needed some exploding barrels or random crates around the place and it's a charmless FPS.


Imagine this thing coming at you at full speed.


This is an interesting combination of wheels and more articulated actuators (er, limbs?).

It certainly solves most of stability problems for a typical inverted pendulum configurations (cornering, obstacles of different height, lowering weight for more speed, raising it for more reach).


I'm sure this company has no issues with recruiting.


Good point, but they also need the people who really knows both hardware and software and probably a lot of physics too.


News from the future:

Man in a wheelchair beats everyone in a street fight.


I think at some point we're going to have to have another uncomfortable conversation about whether cybernetically enhanced (say ex amputees) humans should be allowed to compete in sports where their new physiology is giving them a distinct advantage.

The same way right now we're stuck between biology and political correctness when talking about ex-male MMA or wrestling competitors dominating in the women's side of the sport when most of their physical development happened before hormone therapy and they end up with a distinctly more powerful physique.

Separate bathrooms for cyborgs who maybe will be reviled and feared when they do become a very small minority?



My favorite aspect of this topic is that amputees long jumpers jump on their prosthetic leg.

I don't think we are past the threshold though. People, including coaches of able-bodied athletes are still viewing prosthetics as balancing the odds.

To me the real threshold will be when a double-amputee will break the 200m or 400m World record. I thought it would happen last year but I guess a few more years will be needed. It will happen.


> The same way right now we're stuck between biology and political correctness when talking about ex-male MMA or wrestling competitors dominating in the women's side of the sport when most of their physical development happened before hormone therapy and they end up with a distinctly more powerful physique.

I happen to be friends with a woman who is in an adult roller derby league. There is a transgender woman (previously a man) who pretty much dominates the league, and is clearly a faster skater than the other women. I'm always wondering how the other women feel about that, because as you say, there is the force of political correctness that stops anyone from outright saying anything about it. But it's clearly an unfair advantage for the transgendered person.


I think this is the most frightening company there is.


If I was writing megacorp scifi I'd have them grow via military contracts and eventually merge with Palantir (and maybe some innocent litte AI startup that gets gobbled up, where the hero works) to be one of the megacorps that essentially becomes a government entity (DynamicPalantir). Some sort of Bio/Chem merger (maybe Johnson & Johnson+Bayer/Monsato as a base) would be the counterpart battling it out with DynamicPalantir because they're betting on clones not robots. Toss in Saudi Aramco++ for some good measure.


So what you end up with is an Autonomous, Intelligent, Bipedal robot fighting a clone, eating only GMO-corn, with silky smooth skin? ... Talking arabic... XD


This is really cool tech, but given the current political state, I can't help but worry about a powerful army of robots -- presumably with weapons attached -- being put into the hands of a leader we might consider to have "unreliable" judgement.

I wonder: if a country can deploy an army of robots to achieve its military ends, how does that play out in the long run? Does this mean future wars will just be "our" droids versus "theirs"? Does this "solve" terrorism in the long run, or make it much worse. Also, seeing aggressive robots feels scary.


In the short-medium term? It makes it easier for democratic countries to wage war, just or unjust. No more footage of coffins draped in flags coming off transport planes. No more soldiers coming home with PTSD or concussion-induced dementia. Just a bunch of guys in Langley, sitting in their cubicles, killing people on the other side of the world.

The domestic propaganda goes down much smoother at this point. After all, who wants peace when we can win?


> Just a bunch of guys in Langley, sitting in their cubicles, killing people on the other side of the world.

I bet the lag would be terrible.

Remember Grace Hopper.


> I bet the lag would be terrible.

If they could route the link through a constellation of LEO satellites, then the lag would probably be comparable to what you get in a typical multiplayer FPS. Ground to LEO and back is 1ms to 14ms, depending on how high in LEO the satellite is (assuming satellite directly above your head). Space is pretty close to us.

As for how to get that constellation up, well, SpaceX is working on making that pretty cheap.


I'm actually not from the US so hadn't heard of her until now. But I just read about her, and it sounds like she was a pretty awesome lady! And now you've got me curious about the lag...

I'm guessing they could (would?) go most of the way via undersea cable (just as normal folks do), pop out somewhere close and then beam up to some satellite. I'm in Australia, and (I think) Puerto Rico is roughly on the other side of the word (although I did get a D in high school geography, so there's a good chance I'm wrong here). My round-trip ping time averages ~320ms to some random PR ISP (209.91.222.104). I have no idea how much extra latency a last-stage satellite hop would add in, but let's say it triples the latency.

Almost a second in lag. So pretty terrible. I guess they'd need to use characters, er drones, with good AoE attacks...


> I bet the lag would be terrible.

I'd guess there's no anti-aimbot code, though.


Even if the current policital leadership was someone you considered reliable and trustworthy, you should still be worried, because four or eight years from now someone you don't trust could be in control.


I guess I'd worry more about rogue nations developing (cheaper) biological weapons and chemical weapons. I imagine it would be easier to build something like [1] than a robot like this.

[1] https://en.wikipedia.org/wiki/M139_bomblet


States have always been constrained by what their soldiers are willing to do for them. It's the difference between Tiananmen Square and Russia a couple years later. I think taking away that constraint is a bigger deal than how robots might change the nature of wars between states.

This particular robot doesn't strike me as especially military, though yeah, it's kind of scary-moving.


What is even the point of having a war once you are fighting with robot armies? Why not just throw a coin and hand the country over to the side that picked the correct outcome? Use an unfair coin to correct for military strength. Or each side sends its best players for a game of Risk or Rock, Paper, Scissors or Command and Conquer.

You get a winer and a loser, it does not take years and costs a fortune and you avoid all the unnecessary loss of life and the damage you have to fix after the war. That seems like a win for everyone involved. Well, that could actually also work right now, no need to wait for robot armies.


This was essentially the plot of an original series Star Trek episode, https://en.wikipedia.org/wiki/A_Taste_of_Armageddon:

  In this episode, the crew of the USS Enterprise visits a planet whose people
  fight a computer-simulated war against a neighboring planet. Although the
  war is fought via computer simulation, the citizens of each planet have to
  submit to real executions inside "disintegration booths" to meet the casualty
  counts of the simulated attacks.


You get rid of destroying assets but keep killing people. It is kind of an improvement if you just compare the losses incurred in this kind of war as compared to normal war, after all you are not losing your assets. But morally it seems actually worse to me. You decide to lessen the consequences of a war but you decide to keep your assets but keep killing people.


It won't work because actually enforcing a meaningful loss requires a real war.

Otherwise, no matter if you believe if you have got a 50% or 10% chance of victory (or even no chance, but a certainty of painful losses for the opponent) then it makes all kinds of practical sense for the loser to say "screw the coin/game" and simply ignore the result and escalate it further.

If a reasonable settlement acceptable to both sides can be negotiated, then it doesn't need the coin or game in the first place; and if someone can enforce the conditions on the loser without a horrible cost, then there's no need for a game or a war since it can be done as a simple ultimatum.


For lack of a better word, that was meant more cynical than something that has actually any chance of working. It just seems to be a somewhat logical continuation of the idea of a war where everyone just sends robots to the front lines to fight things out. But you are of course right, the correct conclusion is probably that there can not be a war just fought by robots.

After one robot army has defeated the other one it has to go after human soldiers, assets or even civilians to achieve a victory or must at least be willing to do so unless the opposing country surrendes at this prospect. Just defeating the robot army and then doing nothing will not be effective beyond the inflicted loss of resources.


If the simulation is reasonably predictive and the spoils of war not too severe then the loser won't start the war because they would npt expect to gain enough.

This is what democracy is, a coarse estimate of who has the most soldiers. A 90% majority can do what they want, a 51% majority needs to compromise a little.


Just as the purpose of a war isn't to determine who's stronger but an extension of diplomacy with violence as a tool, a simulation isn't needed to determine who's weaker, countries already have a good idea on that. If the results are reasonably predictable and the request is not too severe, then you don't need a simulation, game or coin toss, ordinary diplomacy does that well; this is how events like Sudatenland takeover in last century or Crimea takeover in this century happened without a fight.

However, in an actual war many things matter more than the numbers or fighting power. For example, it is clear that ISIS is much weaker than the full USA military, however, because defeating ISIS in an actual land war would cause significant casualties to USA, they can go on - a clear simulation result showing that they would lose won't ever be enough for them to concede defeat since they can (and will) simply call it as a bluff and claim that USA is not willing to pay the blood price. So you can't settle a conflict like this with a simulation; the loser always has the backup plan of telling "make me, if you dare" instead of accepting the loss.

Agreeing to a result of simulations that both sides believe isn't settling undecided disputes nonviolently, it's just old fashioned diplomacy or ultimatum/surrender.


If one side has the robot army, they'll target the human soldiers on the other side. If both sides have robot armies, they'll probably target one anothers' civilians.

Nations are never going to rid themselves of the desire for war, or the need for war to involve bloodshed, because we're just another species of simian bludgeoning the out-group to death with extremely abstract and over-engineered rocks. We already have non-violent ways to resolve conflicts between nations, we fight wars explicitly to kill people.


[...] they'll probably target one anothers' civilians.

Which would be a first class war crime.

Nations are never going to rid themselves of the desire for war, or the need for war to involve bloodshed, because we're just another species of simian bludgeoning the out-group to death with extremely abstract and over-engineered rocks.

While that is not an uncommon point of view empirical evidence suggests otherwise, war seems to be in decline and seems to have shifted from wars between nations to civil wars and asymmetric wars.

Some argue that advances in human behavior heavily lag behind technological advances and I tend to agree at least somewhat, but concluding that we will never manage to overcome this stupidity seems not really obvious to me. And I certainly hope that it is not true.

We already have non-violent ways to resolve conflicts between nations, we fight wars explicitly to kill people.

Force is the ultimate and only way to »resolve« conflicts if all other means fail. But I do not think we are killing people for the sake of it in wars, it is just a means to an end.


> Which would be a first class war crime.

War crime convictions are for the losing side.



Is there still going to be a "theirs" standing after a first power deploys massively "our" robots?


Allow me to ask you a rhetorical question: Do you see many Raytheon engineers telling the front lines who and where to shoot?


The moment those drones are given a degree of discretion in target selection, yes.


Because killing thousands upon thousands of innocent Afghanis, Iraqis, Pakistanis, Somalis, Yemenis and Syrian civilians so far with relatively conventional weapons has been reliable judgement? Maybe with robot armies they could have killed even more. These partisan trenches are blinding so many.


Of course those deaths are horrible. Even a single needless death is the most tragic thing in the world. The point of my question was that "conventional" weapons are still limited compared to robotic warfare / AI. Very strong military in the hands of a repressive regime has always been bad, but I fear the robot military age will make it far worse.


Am I the only one who thinks these robots look like Michael Jackson doing a toe stand dance move?


The ghost from Howl's moving castle: http://www.weirdwildrealm.com/filmimages/spiritedaway.jpg


That's not from Howl's Moving Castle. (Clue: The link.)


Good point, thank you. *Spirited Away.


Yes.


Great some one has made my murdering Segway nightmare a reality. Thanks Boston Dybamics.


Much creepier version of this, video of it was actually surfaced about a month ago. [1]

[1] https://youtu.be/giS41utjlbU?t=52s


Heh, last time I missed the piece flying off the robot as it landed the jump. I wonder what broke?


Nice video.

Does any one know if they are planning to make a low power and light exoskeleton kind of mutation to this tech? This could be revolutionary for people with disabilities.

or as a mode of personal transport in difficult terrain?


It looked like it was near capacity with the 100 pound lift test.

I also wouldn't bet on the run time being real long.

The iBot was a practical realization of a stabilized, powered chair (same tech as Segway):

https://www.youtube.com/watch?v=7fY5NYCVvqs

I guess it isn't manufactured anymore.


Now if it just yelled "I'm free!" during that last jump :-)

More seriously, this is a very agile robot I am very sure I would never want to be in the same room with it when it was being tested. That said, the 100 lb basket it a bit disappointing. I was thinking "Ok this could be an offload/onload robot for freight" but maybe not. It is remarkably quiet which is something the original Bigdog discovered really put the damper on its threat ability.

But if Amazon used it in warehouses? Probably more than they need for that activity.


Something human-sized and untethered which is capable of lifting a 40kg load and running with it? It's not going to replace a container crane but (as you mention) it'd be a very versatile warehouse bot.


It reminds me of nothing so much as E-102 Gamma from Sonic Adventure.

MODE CHANGE.


It seems robots jumped directly from baby steps to roller skating...

Add a quadricopter on the head and you'll have a crazy kung fu master...

Freaking cool :-)


Each one of these videos from Boston Dynamics is a little bit cooler and a little bit (or a lot) more scary than the last.


Couldn't this will be very effective in the delivery business we see, amazon, postmates and grubhub undertaking in urban areas?

How about single -seat self driving mobility device https://www.youtube.com/watch?v=rukmDZc3EiA


Why do we always see the BD robots in real videos, and never in simulations? I mean, the mechanics are important, but the control system is much more important, and it's pure software.


What technologies do they use to do the balancing. I heard earlier that they don't use any machine learning at all. So what do they use to achieve this level of robustness.


The balancing stems from an inverted-pendulum problem, and that's something that's reasonably well understood in robotics https://en.wikipedia.org/wiki/Inverted_pendulum

Of course it's more complicated than that because it looks as though the centre of mass can move and the legs can bend and are not rigid, but it would make more sense approaching this from a kinematics perspective rather than a ML one.


I was thinking of the Wheelers from "Return to Oz": https://youtu.be/2MLVfQOknYM


Driving past their office has got to be a weird experience.


Mankind's days are dated. Couple this with AI and we'll have a whole new species on Earth. Start building your own bots and restock your underground shelters.


I wonder what it's like on a halfpipe :D


Otherwise known as Parkourbot


Get that thing away from me!


wow this really has the potential to revolutionize the adult film industry


Impressive!


That jump!


does it go up stairs as well as down?




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: