How long before we have the YouTube remixes of this video with horror film music and a jump scare in the end blackness after the Boston Dynamics logo fades away?
Is feels like they almost designed this video with that eventuality in mind. Like that wanted a second wave of organic 3rd party viral advertising.
As usual, science fiction predicted everything exactly the opposite. It was thought that robots would handle hard physical labor while humans would engage in creative work...
Making war feel less risky to politicians who want to wage it for domestic jingoistic bullshit reasons[0] makes war more likely to happen in the first place, which costs lives.
Also, bold of you to assume you're going to be winning. Does the excitement about this new tech hold up if you consider it from the perspective that it's going to be used against you and your troops?
[0] as opposed to actual defence against invasion - despite the euphemism commonly used by western governments for their military political departments. And what %age of military actions in the last 50 years that your country was involved in count as one, or the other?
You're making so many assumptions that you're attacking a scarecrow. I think that more precision (which may make war "easier") makes up for it in total lives spared. How are militarized robots worse than a JDAM? If I didn't believe that it is more merciful on the whole then I wouldn't take this position.
What I find funny is just like with generative AI, this was under the Google banner first until it got struck with office politics related to Andy Rubin. I still don’t understand why someone else at Google didn’t take it over. They really lost their way a long time ago.
So as robots became more capable and less contained (like current industrial robot arms). What novel tools/techniques there are to stop misbehaving robot?
Will casting a net stop robot like this or do you have to somehow dismember it?
These robots will be available to buy within our lifetime. Look at the nicely designed power button on the back with blue light ring. They are preparing for consumers.
An elderly relative of mine has smoked his whole life. He admits it's bad but won't stop regardless of pressure from family. He enjoys the routine and the quiet moment. He would absolutely send his robot to buy cigarettes if possible. Nobody else in the family wants to buy them.
Recently got a tour through boston dynamics, but mostly saw the spot department, the atlas department was off limits. I guess this was the reason then :) very cool
Good, the hydraulic version sucked ass. Jerky, unpredictable power delivery, when it wasn’t broken. Was such a pain to model and design kinematics around.
No, I wouldn’t. And I appreciate the Boston Dynamics track record. That said, they do have a reputation for viral video releases. Curiously, they usually show some sort of human interaction with their new robots, but maybe too early here? Particularly loved the video of the engineer shoving the robot as it negotiated an obstacle course.
Would you bet money that it isn’t CGI? The fact that we are asking these questions is just as terrifying and impressive as the malicious potential of the robot itself. Wild days we’re living in!
Their marketing style is built around viral Youtube video showing their prototypes being impressive or simply entertaining. Including when the fails.
They got a lot of recognition and attention thanks to this, I would not blame them, I prefer this to sterilized marketing we're seeing most of the time.
I think the parent and gp poster are more like, "Tesla is an odd reference. Not wrong per se, just odd."
It's like saying, "they are building a search engine, just like Netflix!" Sure, Netflix does build search, but like... are they the canonical example for the domain?
They did stress that one advantage of electric motors over hydraulics is better mobility. On the other hand, the motors probably do not yet have the power to make jumps and the like.
I'm not sure that when the robot overlords look through their training data if they will decide that "huytersd" was being serious of facetious. This may not have the effect you were looking for.
This guy seems less creepy than his predecessor. He looks more like a hobby servo-motor robot. I liked the aesthetics of hydraulic Atlas better - somehow fitted well with the character they gave him in all the choreographed demos.
I can't see them really being creepy unless/until we get to "uncanny valley" territory with realistic faces and expressions.
Jesus fuck. I guess the war machine is hungry again so they've fired all the people who made the cute dancing videos and brought in the nightmare engineers.
I'm thinking more and more that that "Terminator" was the most accurate of all the sci-fi dystopias.
the war machine is never not hungry, Boston Dynamics has received some funding from the Department of Defense and has sold robots to various police departments and other government agencies
Really scary feelings watching the clip.
I think we need to make robots either neutral, or somewhat cute. Otherwise society will distrust these entities. This is the opinion of someone strongly rooting for the success of AI/ML and its symbiotic integration with actuators, either on an isolated basis or as a large hive mind.
Does anyone remember the scene in Terminator 2 where the T-1000 turns around instantly by swapping its face from front to back on its head? It reminds me of that. It's like they were consciously trying to evoke the Terminator.
Misplaced empathy. It isn't some disabled people like on a wheelchair, but people suffering from brain damage that makes them terrorize their surroundings. You can hardly find a better solution than to make places unbearable to them by intentionally triggering their hallucinatons.
A hill I'm willing to die on: bipedal robots are an evolutionary path that machines don't need to go down, we have lovely bearings and wheels that work perfectly with electric motors.
Yes obviously there are limitations i.e. stairs and uneven terrain but there are wheeled/tracked solutions for those too
Most of these robots will be used in factories that have very nice flat concrete floors
The ultimate goal is to produce general-purpose robots. If we want robots that can do everything a human can, then legs are definitely useful.
One simple example: getting in and out of a car. Another thing to consider is that a legged robot can tilt itself for balance while carrying heavy objects. To carry a similar weight with a wheeled robot you'll need a much wider wheel base.
And then of course, if you want to build robots that can be useful inside a house, then they need to be able to cope with stairs. There's also construction... At some point, you don't have elevators... Or just circulating between buildings out on the street where the pavement isn't great.
In my opinion your argument assumes there would be a single form factor for robots that will be used everywhere. This assumption has generally been false for most technology, look at the different cars or personal computer. In my opinion, we will have as many kinds of robots as there are breeds of dogs, some of which will be bipedal, but most of them will make do with wheels.
Exactly. We don't need robots that replace humans 1-for-1. If there's a building site that currently needs humans to scale ladders etc then a combination of lifts, loading bays, cranes, drones and tracked robots can do it, not legged robots that carry everything up ladders etc.
Of course that needs very smart systems that can co-ordinate but that's my point, there's an opportunity cost for everything, and I think that's better spent on AI and a multitude of other systems rather than a schoolboy sci-fi fantasy of bipedal robots
> look at the different cars or personal computer.
this feels like a flawed example to me. ~100 years on and all cars are starting to look the same.[1] Personal computers, after like 30 years, have mostly converged around something that's essentially a 3x5 touchscreen with cameras on both sides. Sure, there are laptops and PC's, work and semi trucks, but that's 3 form factors? meh. Manufacturing at scale is much more efficient, and form follows function, can't really escape either.
To me the killer app would be basic house chores: cleaning, doing dishes, etc. For industrial applications I suspect we will retrofit factories for whatever robot tech we have, instead of needing humanoid robots to use interfaces designed for humans. The same goes even for commercial applications like stocking grocery shelves. Driverless cars are an obvious example already. But people probably don’t want to significantly retrofit their homes with less human-friendly interfaces.
Why do I want a general purpose robot? I don't need one robot that does it all, I'm happy with a separate robots for washing my dishes, and vacuuming my floors. Sure both can be improved on, but they don't need to converge. In fact I'm glad those two are separate as I can let both of them run at the same time and get the work done faster, while a general purpose robot can only do one at a time.
The goal is to make my life better, robots are only an implementation detail. Maybe some robots need legs (construction robots?), but most don't. If the robot with wheels is cheaper I'll take that in many cases.
I don't think you want a vacuuming robot. Those already exist, its called a Roomba, and they have a lot of limitations that are completely intractible in their given form factor. You have to modify how you use your house to make it actually useful enough. Some examples:
1. Stairs. Roomba's can't vacuum stairs, so you still need to do those yourself.
2. Stairs, Roomba's can't traverse stairs, so you need one for each floor.
3. Doors. If you want the Roomba to vacuum the whole house, you have to have all the doors open for it.
4. Can't have anything on the floor, the Roomba will either get stuck or avoid it. But I shouldn't have to never leave a backpack on my floor if I want it vacuumed.
5. Corners. Roomba's can't vacuum in corners or in tight areas between furniture and walls. or any other weird geometry. ie: I have a wire shelf. Roomba doesn't fit under it but its easy to use a stick vacuum to get between the wires and to the floor.
And this is before we get to the limit on suction and capacity in that form factor.
The roomba isn't the only possible form of robot vacuum. There may be other options for a design that isn't general purpose but eliminates those issues. Perhaps we should just install elevators or dumb waiters in houses (dumb waiter may be cheaper because it doesn't have to be human rates for safety, while elevators are also useful for humans in wheel chairs which at some point in your life is likely to be someone you are close enough to that you would want to invite them into your house). Likewise doors that can open themself are an option that can solve other problems (think star trek - not current technology)
You want a separate robotic appliance for every thing you want done? That sounds... hectic.
I'm also not sure it's important that laundry and dishes get done at the exact same time - if it is, you should probably do 1 of those tasks yourself - especially since a robot would be able to stuff at night, etc, giving it more time to complete tasks
Also, GP mentions stairs, and adding wheel support to that. So not only do you want a half dozen robits rolling around your home, you'll also need to remodel your home to support it.
Or, of course, we could develop bipedal robots, which seems to have little downside as compared to wheeled robots.
I just want the tasks done without thinking about them. How it happens doesn't matter to me, just get it done. 1 robot, 1 million - I don't care, just so long as I can afford them and they stay out of my way.
While I don't care if everything gets done at once, I care that things are done right and not otherwise inconvenient for me. Maybe the best way to have robots that work slow and then apply a lot of then.
The important thing to note here is robots for many of the things I want do not exist. When they come we will see. Maybe is a a specialized robot, maybe it is more general purpose. That is irrelevant.
My phone is not 100% general purpose, it is a compromise. I'm typing this on a computer not my phone that is right next to my keyboard because my phone is a bad compromise for typing comments. The phone works well enough that I'll use it when on the go, but only because there are times when hauling my full size computer isn't a good option, as soon as I'm using it I prefer the computer.
There is a lot of room for special purpose tools to handle more than one purpose while not being fully general purpose. I'm suggesting we never have a need for full general purpose, but there is for sure room for robots that do more than one thing but don't do everything. I might want the robot that sets and clears my table after meals to also gather my dirty laundry and when clean bring it back - but offload the actual cleaning process for both to specialized robots.
That tilting wheel robot is huge compared to a person, and it seems to me that the potential for it accidentally injuring a person while moving would be much greater than with a humanoid, if only because of its mass and its need to perform large, rapid motions to maintain its balance.
I think we very much do. Robots are currently very expensive, so where do you want to send a robot that you can't use a worker? Probably somewhere at least potentially dangerous.
You want to use the robot to inspect a tunnel in danger of collapse, or a factory that may be leaking a poisonous chemical out of a pipe.
And in such cases you very much want something that can navigate obstacles about as well as a human. You can't count on the area being devoid of rubble, and rebuilding a factory to make it wheeled robot friendly could be an enormously expensive and impractical proposition.
Now humanoids? We already designed everything for us. A good enough humanoid robot can go anywhere a person can, and manipulate anything a human was intended to touch.
I would invest in the spider-legged robot to crawl around spaces.
I think human physiology is amazingly multi-purpose, but we don't need to compromise on balanced skills with robots. Every action can have a physiologically tailored robot to do it. Sure, I can see that I would want my personal butler bot to be humanoid, but I think for the vast majority of cases, humanoid is not the optimal solution.
But I also suppose that if I was going for wooing the general public, I would go humanoid for sure. People compare technology against science fiction, not actual practical considerations.
But a humanoid isn't the optimal form factor to be able to navigate those kinds of terrain. A quadruped robot like Boston Dynamic's Spot is much more stable than a bipedal one, and is already being used for those sorts of applications.
For rougher types of terrain, hexapod robots do great (not the spider-type ones - ones with three legs either side, that fully rotate in the vertical plane), or for that matter just use a tracked tank-type design.
A combination of a tracked vehicles and drones then. There's something quite short-sighted and uncreative about assuming bipedal 1-for-1 replacements are the only solution
Agreed. It seems to go hand-in-hand with people wanting to demonstrate humanoid robots doing domestic chores like shirt folding.
I'd go out on a limb and say that we will NEVER have humanoid robots at home folding laundry, walking upstairs to put it away, or putting away the dishes in the kitchen. This is a 1960's sci-fi vision of the future, similar to that of flying cars. Any robot capable of fully navigating the human world will always be too expensive and unreliable as a home helper.
In a factory a stable wheeled robot is way more practical than a bipedal one. It doesn't need a humanoid head either - but I guess that makes for nice PR photos.
The problem with that “1960s vision” is thinking to literally about having robots do exactly what humans do now. Likely there is a creative way to solve the need for humans to do those tasks via automation, but it’s not likely to look like a humanoid robot folding laundry.
On the other hand, it'd be highly amusing if the future did involve humanoid robots out mowing the grass with a push mower, or getting into their car to drive to the grocery store.
I'm reminded of this Adam Savage video of BD's Spot pulling a rickshaw (23:00).
Humanoid robots making buggy whips to control ornery LLM-based robotic horses. With private/pair key encryption in the whips themselves- they can send a digitally signed wireless "threat" before actual contact is needed.
I don't think its too outlandish to see these getting to a price point where they're cheaper than a human. The goal might not be an appliance Rosie the Robot in every household but having a robot that can help the infirm, elderly or disabled.
If we ever see a world where robots need to be useful outside of a factory with perfectly flat concrete floors, then yes - there needs to be continued evolution in traversal over uneven ground and around unanticipated objects. Bipedal locomotion is useful for this (although not the only solution).
Right now the hardest jobs to replace will be those of plumbers, electricians, carpenters, etc where they need to operate with fine motor skills in unique and challenging locations - no two ever being the same.
i think search and rescue is a great application of humanoid robotics, you need something very versatile and a human body is not a bad model for a universal terrain form factor.
What about when a robot is carrying an uneven load and has to rebalance?
What if it is knocked over and needs to get back up?
What about Steep inclines? Stairs?
What if it needs to climb on to a different platform? A conveyer belt? A vehicle? A beam?
Even in a factory or warehouse setting wheels are useless for anything but the most ideal cases. And there are already countless robots successfully operating in that space. A general purpose robot is the holy grail, and legs are a requirement for that.
I think it really depends on where the robots will be used. yes short term they will be in factories, shipping centers, etc. Places that can be tailored to the robot.
But the long term prospects of robots would be in your home, maybe going to the store for you, whatever. We see the limitations of wheeled robots with robot vacuums. They do a decent job but are severely limited trying to do its job in a place that was designed for a human. (On the flip side it can also get some places easier than a human would, so it's a bit of a trade off).
By focusing on mimicking humans, we end up being in the best situation for both of these. Factories can try them out with minimal changes to how they operate.
Plus, it seems like the biggest hurdle isn't really walking. It seems like we have gotten that one down fairly well (not perfect obviously) and the bigger issues seem to be hands, object recognition, and just "general" AI. Can it actually do anything with the hardware it has on its own.
Boston Dynamics makes all kinds of robots. None of them are consumer products, so I don't think the very good arguments, like Angela Collier's, against having one in the home will be an impediment to developing very capable humanoid robots.
Seeing how this one moves, it is human-ish, being bipedal, but it isn't mimicking human movement range.
One of my pet peeves is the idea of asking the world to accommodate a situation rather than build solutions that adapt to the world.
Big example: the best we have for mobility nowadays is a wheelchair of some sort. That requires building special ramps and elevators everywhere.
If we had a four legged chair that could climb stairs, etc, like what BD is doing, it could transport people ANYWHERE. you could literally go for a stroll in the woods with it. People that are injured for 6 weeks in their home could go up and down steps, etc. The elderly could go for walks in a park.
So I for one fully support more research into smarter mobility that doesn’t require the world to accommodate it, but instead adjusts to its surroundings.
Segway make wheelchairs that are far more versatile than traditional ones. I'm not sure of the capabilities of their current commercial models, but years ago they had demo videos of them driving up steps and a scissor-like design whereby they could lift the occupant up to reach things.
> One of my pet peeves is the idea of asking the world to accommodate a situation rather than build solutions that adapt to the world.
While I understand and respect the sentiment, in my opinion human history has been a trend in molding our environment to our advantage. I can drive to a remote hill in Bangladesh from the capital because there are roads that we humans built and maintain. If we kept molding to the environment, such an accomplishment would never be possible.
So yeah, maybe mold to the environment a little bit, but also mold the environment a bit, is the ideal solution.
I think this is an absolutely horrible thing for the environment. My point was to make a testament to the human will, and also to counter arguments that try to wield the cost of accessibility against people with disabilities. We could make a global network of ships, planes, and cars, so why is making a tiny ramp such a big deal afterwards?
Not really, not fast nor convenient. Any machine will always add extra volume and weight in the most inconvenient ways. There should really be no limitation on the designs, just optimization under the constraints at hand
it's hard enough for disabled people to get a non-shit wheel chair, you think the world
is going to give them the most advanced quadripedal robotic walking system of all time for nature walks in the woods?
What changed my mind is thinking of humanoid robots as the “last mile” of robotics. All the thousands of use cases where there are no easy patterns and we need something that can fit into any human task without planning or modification.
I'm pretty sure human's role, in the grand scheme of things, is to generate the next step in evolution. We should do the best job we can for the universe.
It was mostly tongue-in-cheek, just to offer a different perspective. But since you ask, how should we employ our capacity to care? On fleeting comfort, or grand visions? Personally, I vote for creating a superior life form, that can carry on the long history of evolution into amazing new realms and abilities. We could be the "bacterial" precursor of an amazing new stage of evolution.
Once they get walking wit legs perfected, they can install wheels on those or do whatever wheel thing they want. That will probably be an easier addon.
I too think it's a distraction too but it's won't be the limiting factor. Planes don't flap their wings, cars don't have legs yet are faster and more powerfull than animals.
The important part lagging is the brain. Understanding the world, reacting to it learning. Even an ant can navigate the world pick-up objects and do tasks.
Well yes. There's always a robot that can be specifically made to handle a specific task in the most efficient way possible.
But the fact is, the world has mostly been built by humans for humans. Pretty much any task you can think of can be accomplished by a human with their arms, legs and some tools.
Androids are human-compatible. An android could go any place a human could go and operate any machinery a human could operate - that widens the space of possible applications. A wheeled robot is capable of many tasks, but it can't dance with you, play piano, wear your wardrobe or sit in a plane seat.
I think you're unnecessarily short-circuiting your imagination.
For one, there are many applications in dangerous environments that could benefit from the dexterity and ability of bipeds - rescue missions, mining, space walks, etc.
If we’re using NN’s to get fine motor skills right, like final steps in an assembly line, the simplest and most abundant source of training data are humans. :shrug:
Sure, but you don't even need a lower body for that, and if you do want to let the robot move around than a stable wheeled base that doesn't negatively impact the fine-motor skills needed when it is in position seems preferable.
Boston Dynamics: Hey everyone, we're really excited to show you the great progress we are making in our attempts to re-create the Torment Nexus, from the classic cautionary sci-fi novel Don't Create The Torment Nexus.
Too many responses: Oh, wow, it's so creepy, just like the book! Lol. Anyway, I'm pretty sure it won't turn out as bad as DCTTN. ;-) Best just get on with my day and mostly forget about it then...
Using my human brain, I can't tell if this is real or not. One day the robots are going to watch this video and decide they've had enough, and that's how we all end, joke video or not. I really do hope it's a joke, but there's not enough money in the world for me to start smacking around a robot holding a live firearm.
