531
Boston Dynamics introduces a fully electric humanoid robot that “exceeds human performance”
(spectrum.ieee.org)
this post was submitted on 18 Apr 2024
531 points (96.7% liked)
Technology
59161 readers
2119 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Okay somebody probably knows better than me, but what is the advantage of humanoid robots? Why are people kinda, on this, now? It feels very 20th century, as an idea. It's pretty cool, but I don't understand why this would be necessary compared to just like, specialized normal robots that do specialized normal tasks. It seems more efficient, if you wanted a robot to, say, do the dishes, to make a robot that just does the dishes, instead of making a robot in the shape of a person that does the dishes. The one that's in the shape of a person is maybe more broadly applicable to human contexts, software notwithstanding (which does seem like a major hiccup). But it's not as though there's like, an upper limit on the amount of robots which we can have, in total. You could just make more robots, and make them specialized for certain tasks, like stocking shelves or whatever, and that would probably be easier, I would think, than making one robot to rule them all. Like, one robot, with ostensibly an on-board computer and on-board batteries so it's as universal as possible.
It gives me self-driving car vibes, where we could've had them in the 50's if everyone was willing to install metal spikes in the ground every however many feet ahead, but then that maybe doesn't make any sense, because it's just kind of a shitty train or tram. Basically, that nobody's willing to front the cost of infrastructure for anything anymore, so we have to make like, a universal device, and end up quintupling the total cost while making a solution that is either less efficient or doesn't exist.
Also, what's the point of the legs? Is it supposed to go outside, or go up stairs better? We already have pretty efficient wheeled vehicles capable of doing that in most public spaces, or, we're supposed to, anyways, they're called wheelchairs. What do we need this guy to walk around for?
Edit: So far, what I've gotten is ladders, and the scalability of a humanoid design vs other kinds of designs.
For ladders, I think you could probably tackle that with a similar set of constraints what you might need for a stair climbing robot, maybe just with a couple heavyweight anchors inside of it and some gearing or something. Use the robot's big arms, the manipulators, to climb like normal, and then use the bottom wheels to sort of ratchet the robot up. Probably that could work on most ladders with some clever engineering. Could maybe also run a cable from the top of the ladder to the bottom and then have a system where the robot rappels up and down for lots of ladders, but yeah.
I don't think you end up spending all that much on a robot that has wheels vs legs, and I think probably the increased efficiency would be worth it if you want like a generalized robot here. Might be wrong, maybe a roboticist can tell me no, but I dunno. As far as engineering goes it doesn't seem any more complicated than legs. Legs seem better for, offroading, basically. Which are why lots of animals use them, cause animals don't have roads.
For the versatility and scalability thing, I dunno whether or not it's more or less efficient still. While a steam cleaning room does require some amount of consideration to build, I would think that you could really get the price down from the tens of thousands of dollars required for this kind of robot to perform a similar job. Especially if you built it that way up front, retrofits might be much more expensive considering how many bathrooms aren't built correctly. Or you could go with a kind of hybrid approach, which I totally forgot about, but would seem to make some amount of sense, especially for a larger building.
Maybe those savings build up over time, and you could just have a McDonald's staffed by like three of these at once and only spend like 500,000 on it, which does seem like quite a bit to add on top of a McDonald's opening costs. I'm assuming you gotta buy multiple to staff it as a whole and also that you have to buy multiple to have more battery capacity, but maybe one of these will come out with the clever innovation of a swappable battery if/when they come to market. I would definitely hope that'd be the case.
I'm not sure it works out economically. I'm not sure of anything here. These are just suggestions because I haven't seen a lot of FUD on these human robots, except of the Terminator variety, which seems dumb.
I suppose my biggest difference here is just one of philosophy, mostly because I've seen it reflected in self-driving cars. You can make something that's capable in any context. Wind, snow, rain, shine, heavy traffic, pedestrian traffic, intermodal traffic, different kinds of roads not created to a set standard by the state's DOT. You can make a Swiss army knife, right. Maybe there are some economic and QoL savings there if you can do rideshares or do like, johnnycabs, right, if you can eliminate the desire for car ownership and status from the American mind. Maybe you can get all those cars out of parking lots as much as possible, and onto the road, doubling, maybe even tripling the amount of traffic as cars now move from one person going from one place to another place. Maybe you can also solve traffic, if you can get all the human driven cars off the road and totally automate everything so none of the cars ever hit each other or anyone, maybe you could try to do this piecemeal with autonomous vehicle only zones and surmount the nimbys with venture capital to buy a whole local municipality. Maybe you can balance the speed with the safety so we don't have heavy traffic and we get places on time, but when a disaster strikes from ice buildup on the road in a random place, or leaves flying around your car, it doesn't kill everyone. Maybe, you can get all this to not computationally require an energy cost collectively that rivals a medium sized European country. Maybe you can also solve the wireless communication issues that would plague this system, and maybe that allows you to simplify it instead of having every car be self-driving and predictable even though probably a bunch of different companies will be trying to get in on this and will be mingling in traffic, with different softwares.
Maybe you could also make a tram, though. Maybe you could have an electric folding bike you take on the tram. That's also an option. That's just infrastructural cost, and we could do that right now.
Neither of these solutions is necessarily better, right. I mean, in this case specifically it's pretty apparent to anyone with half a brain that the trams are better quite obviously but like, as an analogue about humanoid robots, especially with the point taken about a variety of contexts as opposed to a single high friction context like cars, neither of these solutions are necessarily better. They entail different philosophies. One created a world around the thing, one creates the thing. One changes the world to suit the people, one changes the people to suit the world.
Is what I'm saying making sense, did I get my point across?
We are slowly, step-by-step heading toward visble androids.
Other people have mentioned flexibility in maneuvering around humanoid spaces, but there's also something to be said about bipedal balance. I can pick up a pack of shingles throw it on my shoulder climb up a ladder and put the shingles on a roof without having to be half of the size of a cherry picker lift. Nothing says they have to be exactly humanoid or even necessarily bipedal. For some jobs I could see something more tri-pedal, insectoid or even spider-based would be useful
Operating in environments made for humans, as well as alongside humans.
Say you have a remote hydroelectric dam that needs inspecting. You might need to walk stairs, climb ladders, open doors. We're neither going to re-build all the dams we already built, nor build new ones to be wheelchair-accessible.
All this R&D on a drone that can open doors?
Yeah, since we've designed our world for humans, the best general purpose robots will have a human shape in order to function effectively in the same areas.
Internet is full of text and video about humanoid bipedal things performing tasks. There's a lot of investment money gambling that LLMs + human-shaped robots can make an electric slave that an unskilled overseer can vaguely yell at and the robot can figure out what they meant and how to do it like a human can.
Wheels need roads and paths. Legs can walk over just about anything.
Specialized robots are great at doing one thing and terrible at doing everything else.
General humanoid robots can traverse and operate in environments that were designed and built for humans, and vice versa.
There’s a reason evolution settled on the tetrapod body plan. It’s very versatile. Walking on two legs just adds to that versatility.
It's easier to build a specialized robot for one task than to create a general purpose robot to handle that task. However, as the technology matures, I think it becomes much more practical to create a general purpose robot that's capable of performing millions of tasks than to create millions of different specialized robots. Not only is that far less to design, source parts for, build and maintain, but it also makes it much easier to repurpose them as needs change. The same basic design can potentially be used for factory work, household chores, new construction, search and rescue operations, food service, vehicle maintenance, mining, caring for kids/elderly/pets, building and maintaining other robots, etc. We're not there yet, but that's where this kind of technology could potentially take us.
The advantage of a mostly humanoid robot is that it's versatile and can use existing solutions built for people. Yes, you could replace the legs with wheels or treads, and you'd probably be just fine for most functions with a Johnny 5 type design, but there will still be exceptions. Being able to climb up or down a ladder for example means that you don't have to engineer a solution to deal with getting onto a roof or down into a tunnel system. We've already spent thousands of years solving those problems for humans.
Exactly. In a factory you can create an automated line that runs as long as you need. Or you can spend 1.5-2x that and get an automated worker that can work any of the lines you already have.
Also a lot of tasks that take one person would take several robots. Modify and inspect part-pack part-move pack to pallet for example
Because if the abrahamic god made us from clay, we, who have the spark of divinity inside of us because we were made on his image, can also do it and will do it better because fuck sky daddy.
War and policing. You're dug into a foxhole/bunker with your bolt action rifle, and a few thousand of these things come marching along. Or you're protesting the party in power, so a few hundred of these automated law enforcement officers get sprung loose to "keep the peace"
You do understand how expensive and impractical that would be, right? Training a thousand 18-year olds to handle anti-material rifles is going to be far, far cheaper than doing maintenance on a hundred of these robots.
You don't need humanoid robots for that.
Also battery life, You can give a human a couple of sandwiches and they'll go all day, the robot probably needs a 20kg battery and still won't go more than 8 hours without charging. Its actually one of the kind of mindblowing things about organic systems, A whole human body runs on less power than just the control unit for one of these robots (assuming its similar to a mid range laptop) let alone the motors.
I have a very simple test for these things when it comes to military stuff - would this or that piece of tech have proved an obstacle to the NLF (you know... that organisation people insist on calling the "Viet Cong)?
And the answer for this robot is... only if the objective was to provide them with lots of fancy scrap metal.
Wait how many Kcal does a laptop burn a day?
Its based on a rough calculation, a humans energy consumption averages to about a 126w supply, a mid range laptop uses about 150w. Admitedly a humans energy consumption is going to change a lot moment to moment but then so is the robots.
Hmm... well considering small sedentary adults need like around 1,200 to survive, then a mid range laptop seems like it would require 1,400.
Nah its based on a 2000kcal/day person, there's a still a lot of caveats sure like the person has to sleep but considering we're comparing to literally just the robots brain a human is way more efficient.
Expensive and impractical today, yes. But we would say the same of every man, woman, and child having a supercomputer with the entirety of human knowledge in their pocket fifty years ago.
We have set up the world to be very usable and traversable by humans. A robot in that form would be able to go and do most things. Think of all the changes to places that needed to be made as part of the ADA. And those were for humans with relatively small changes such as wheels instead of legs.
Because this makes it a drop-in replacement for a person. You need to go into a confined space with toxic gas? Your Wall-E-looking robot works great, right up until you hit a single step, god forbid a ladder. This robot could handle all of that, and turn the valve to shut off the gas or whatever.
You could make a spider bot or take most recent alien designs in TV and movie and it would have better dexterity than a human.
Or a snake bot, obviously.
Here is an alternative Piped link(s):
snake bot
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I'm open-source; check me out at GitHub.
They could make more specialized robots, but I imagine the selling point is the versatility. A specialized robot can make food at a fast food restaurant, but can it also deliver food to customers, mop floors, and clean bathrooms? Adding a specialized robot to a kitchen or a factory floor may require you to completely re-design how the floor/room is set up, but adding a humanoid shaped robot would not require any extra setup (well, besides teaching it/programming it).
Spot has been used as a security guard/ inspector at some sites. Going up stairs/ladders would be extremely difficult for something on wheels/treads, but for a robot that works like a human it could be easy.
I mean I guess my main point is just that I find these robotic humans to be kind of interesting and cool, but also totally unambitious and cynical, precisely because they don't require things to be redesigned in any way. You could have floors and bathrooms designed so that they rinse themselves out and are steam sanitized with the press of a button, and not that much of a change in infrastructure compared to the norm, it's just a different way to doing things that requires more thought out design for more discrete applications. Most fast food bathrooms already have a floor drain. Same with food delivery. There's conveyor belt restaurants, we already have drive thrus with windows, restaurants already need to be navigable by foot in order for people to be able to sit down. These seem like maybe more efficient solutions long term than replacing people with humanoid robots.
Are the numbers working out that these are actually more efficient, economically, or something? Is it just that the tech sector is what has all the money lately, so they're prescribing these as kind of a blanket solution to every other use-case? Is it just that human shaped robots are kind of cool? I dunno, but I do wonder about the versatility being more efficient. It works if the versatility itself, in a human environment, is kind of the end-goal here, but as a good in itself, I dunno.
My dad used to tell me "It's a lot harder to carpet the world than it is to wear shoes."
Ambitious redesigns of existing infrastructure are neat, but they are rarely more efficient or practical. Especially when you are overengineering to solve an issue that's already been dealt with. A self cleaning room requires a lot of additional hardware, all of which has to be designed, built and installed, and has to be powered and run by software that needs to be programmed. It also needs to be maintained, and depending on how it's cleaning things, it may also be dangerous, or at least capable of damaging property (ever have a motion activated light turnoff while in a bathroom stall? now imagine it triggers steam jets). Not to mention the potential hazards of water damage on a room if anything goes wrong.
Or, you can buy a mop for 0.1% of the price.
Humanoid robots can escape this problem because versatility adds value. The upfront cost may be tens of thousands of dollars, but for that price you're getting something that solves many, many problems. They can potentially go from task to task, filling a multitude of roles, and ideally with minimal down time.
It also helps that we can use existing processes to train them. They can observe human workers performing a task, attempt to replicate that task, and use feedback to improve. And that's critical because the hardware is the easier part, it's software that's the real challenge.
I also question why they think it needs to have legs, especially only 2. The dogs seem to have way more stability on 4 legs. Why do they insist on making them appear human?
Because our spaces are also made with pets in mind and that form factor provides a size and adaptive benefit
Gimme the spider designs! I'd be good with both ghost in the shell and wild wild West spiders
Centaur!
Yeah! That's more like it!