Many companies have “humanoid robot carrying a box” or “dog with an arm and “fingers” to open doors and press buttons”.
Isn’t it easier, less complex, more efficient, less energy consuming, etc to make a box-like robot with forks to carry boxes instead of a humanoid?
Is the dog-version more efficient running than a tank-like robot? If that’s the case wouldn’t a spider-like be even better, specially for steep hills?
Comments
Most robots are robot-like robots. Its just that there is more attention ok robots that look like humans or animals.
they already do build robot-shaped robots, and those are highly specialized in their tasks.
> Many companies have “humanoid robot carrying a box” or “dog with an arm and “fingers” to open doors and press buttons”.
these are just the first steps in trying to perfect a multitasking robot that can do multiple jobs, and what’s the best example of an adaptable multitasking machine than ourselves.
Most robots are “robot” shaped.
You’re right that for most specialised tasks natural shapes aren’t all that efficient, and something much simpler is better. Which is why they do that.
The idea of making robots shaped like natural creatures, particularly humans, is to potentially have a general purpose robot, that can do anything we might do.
If you wanted to automate all your household chores you might imagine needing 100 different specialised robots, or 1 general purpose humanoid.
The overwhelming majority of robots are very non-human like. They look like skateboards with arms. They look like soda machines. They look like boxes that don’t move, but move things around inside them (like liquid handlers). The automated room of a car factory looks like something out of a horror show.
But, regular robots don’t take stairs or uneven terrain well. If that type of robot did uneven terrain well, evolution would have come up with solutions for it that lacked legs. Oh wait, it did. They look like snakes, slugs and snails. But robots functioning like snakes or snails is inefficient for machines. So you make some robots with arms and legs.
Legs are better at handling very uneven terrain, so having 4 of them at each corner of the robot makes a nice stable platform able to travel on a lot of different terrain hence the “dog” robots (which aren’t really that dog like).
For humanoid robots in theory they could be useful for industrial settings with awkward spaces, like ladders, cramped gangways, manual valves etc. An examine often given is if humanoid robots had been available for Fukushima they could have gone in and opened manual cooling valves.
But those situations are quite niche so it’s mostly they are easier to imagine a human shaped robot working in your home (so marketing).
Robotics at that level is still in its infancy. No one has figured out what the “best” form is yet, and there is lots of experimentation and trial and error yet to occur. Most likely, what’s “best” is going to vary depending on the task.
The one argument for making humanoid robots is that everything in our world is already designed for humans. Having a robot that can easily open doors and sit in a car and climb stairs and everything else that humans do means that we don’t have to make any additional changes to our environment to accommodate robots. They will be able to function in our world the same way we do.
At the end of the day, the simplest, most efficient design is not always what gets picked. If that was the case, we’d all be driving egg or fish shaped cars. There are other factors to consider, and a good design is always a compromise on many design choices.
Ultimately, the reason we build robots is so that they can do human jobs. As the whole world has been designed around humans, it makes sense that robots mimic ourselves.
It also makes people more accepting of them. If a robot looked completely alien, like a Darlek or something, people would feel anxious about them existing. Which they are already do of course, but a cute doggo robot is much more comforting that a Darlek.
We already got those robots and have had those for a long time. The point is to make robots that can fit into our human world and replace a human/dog and not having to change the world so the robot can function. Its also supposed to be able to adapt to different envoirments. Like some houses might have stairs so it should be able to climb stairs. Sure we can build a robot on tracks and lay tracks but thats not usefull for general purpose. We already have buttons that can be pressed with a finger so a robot should be able to press that same button and instead of having to build an extra thing for the robot.
We understand how humans work, and how many animals also work. We use that knowledge to make robots with the understanding we currently posses.
Most robots are robot like.
Humanoid ones are speculative. It’s a great proving ground for new tech, and I’d imagine it’s super fun for an engineer to work on. There are some use-cases for humanoid robots already though they’re pretty niche. If you want some automation in a place designed for humans it can make a lot of sense, assuming the tech and cost is there. For example, a humanoid robot in a medical environment could work. It’s mission critical to accommodate people and can be very labor intensive.
Spider-bots are likely to make people anxious and wary. People and dogs are familiar and friendly.
The podcast What’s Your Problem had on a robotics expert a few weeks ago that discussed this. He said the difference was whether a robot was expected to do one repetitive task, or multi-task.
He also mentioned the robot arms in car factories. They have one job, let’s say to weld the door onto the car. So they are just a giant arm built specifically to weld doors onto cars.
If you wanted to break these down into individual tasks and assign them to different robots, you could, and then the robots would look pretty different. A robot whose job was just to open the refrigerator and pull out the ingredients would probably look like a claw machine. A robot whose job was just to put food in the oven and pull it out when it’s done could look like a giant arm with a heat sensor.
But then you’d need 50+ robots in your kitchen. Much easier to just design one human-shaped one that does everything, even if it’s not 100% optimized for each individual task.
The reason you believe that is because you’re getting your robot “news” from media, social or mainstream.
Those robots are toys. Astonishingly advanced toys, but toys nevertheless. Proof of concept of what is possible, and test beds for developing what still isn’t.
Meanwhile all robots with real purpose, doing real work, are “robot-like” robots.
Be that industry, military drones or space exploration.
The majority of robots ARE robotic. A single aperture or apparatus or tool designed to do one thing. Or multifunctional arms, but literally just a giant set of motors and rods.
The human looking robot or dog bots are an extremely small part of what robots we use. They get the most media attention due to being the star of movies, however, hence it may seem like we are focused on them more.
Even in the cases more visible to us, such as delivering a tray of food around in a hospital, our current robots look more like a stool or mini shelf on wheels than a human walking. But a human-like robot will be more visually impressive and get far more attention if used in a hospital.
I think you’ve misunderstood the industry based on popular news. Most robots, and I mean probably 99.999% are as you describe. They don’t get news coverage because they are simple and designed to do a very select automated routines. Also, we’ve been using them for decades.
On the flip side, there are obvious reasons to designs more general purpose robot mimicking life. For the “dog likely robots, it’s a good design for fast all terrain movement, which is exactly why they are marketed towards the military. For humanoid robots, our world is desisted around humans and so a functional humanoid form would allow a robot to interface more generally with everything we make. Also, for psychological reasons we interface better with a human form robot. Cartoons and early sci-fi also kind of set the expectations. But while these grab the attention of the news, they are really more like the absolute fringe of the field.
The idea with humanoid/dogoid(?) shaped robots is for them to operate in already existing human infrastructure. Those robot-shaped robots, do exist, however as others have explained they highly specialized and usually require specifically designed spaces to operate in to be efficient. Think of those giant robot arms that assemble cars, you need a pretty large and open space with plenty of room for them to swing and rotate around for them to make much sense, but if your robot is roughly the same size and shape as Bob over there, you don’t need to invest in special infrastructure just to use your cool new robot!
I think the psychological reasons for making humanoid robots are more important than the technical ones right now.
Actual robots used in practice are usually shaped best for whatever they do and are not animal shaped.
However, making an animal shaped robots has two advantages:
Boston Dynamics did a lot of experimenting on different forms early on.
It’s turned out that the “dog-like” robots are really the easiest to produce and get going. The larger “horse-like” turned out to just take too much power to run, although they were beasts.
They had some mini thing that was slow at walking but could jump many meters into the air.
They had some “bird-like” versions which is more like what you seemed to be talking about, where they were optimized for moving stuff around a warehouse.
And of course they had their humanoid forms. These are generally trickier to build and train but they have the advantage that theoretically, anything a human can do, they can do.
The last one is the holy grail. If you can crack it, then you have turned robotics from being hardware-led to software-led. The really hard part that takes massive factories and investments will be done for you. Now all you have to do is write the smart code that will make it do the right job.
It is like when we went from special-purpose computers to general computers. Now two people sitting in a garage could create a Microsoft or any number of the giant software companies today.
I expect that this will happen with robots as well.
Robot shaped robots are specialized. You have the package manager robot to manage packages. The warehouse floor robots to manage pallets and storage on the racks. Office delivery drones for navigating cube farms and delivering mail. You can’t grab a warehouse robot and get it to deliver mail to the cube farms without restructuring the cube farms and elevators.
Dog/mule shaped robots are good for navigating steep, rocky terrain that tracks or wheels can’t handle. Carrying a heavy load cross country to logging camps or oil rigs in the arctic outside of the deep winter ice road season is dog/mule work or aircraft work. Not wheel/track work.
Humanoid robots are versatile human replacement solutions. One robot can cook, clean, take orders, take out trash and do everything a human employee would do. Meanwhile you need cook arms, trash bots, room basis, straightener bots, cashier kiosks, and the system that ties it all together and manages it.
Take a look of like car construction plants. There are specialized robots everywhere. Same for a lot of industrial machinery.
Nothing that looks remotely like human or dog.
Why humanoid and dog robots?
Because we made the world accessible for us, so if a robot should do the same tasks as humans they have to use the same tools.
Let’s make a simple example:
You have a home robot that should do the groceries for you.
You could make something like a shopping card with an arm that you send to the store. How does it get to the store? By car? Drive itself there the whole way?
So either your shopping card robot needs to be able to drive on streets as well, or has to fit in a car.
If it drives itself it basically an self driving car, with less functions. If it just fits in a car you either need a self driving car or a human to drive.
So you have the perfect shopping robot but it can’t really do its job alone.
Now you have a human robot, that can go into the car, drive to the store, buy stuff using a shopping card, go back, put the groceries away..
You already have everything necessary for a robot to do its job, if it’s like a human, sure we are nowhere near that level of independence, but shows the difference.
Well, we’ve kinda shaped our world for humans.
The doorknob on your bedroom door. The height and width of the hallway leading from your bedroom to your kitchen. The height of the kitchen counter. The handle on the gallon of milk.
We’ve shaped our entire world (homes, vehicles, etc) to suit us.
So I think in some cases, a human-shaped robot is probably better equipped to live in a world designed for humans.
That is, if you’re thinking of a multi-use robot. A purpose built robot to do nothing but assemble cars, doesn’t necessarily have to be shaped like a human. But a general purpose robot that might have to operate in multiple capacities and multiple environments, might work better if it was human-shaped.
The world is build for human shaped humans.
Robots need to be small enough to fit through doors, turn doorknobs mean for human hands, be able to walk up stairs, be able to hold tools meant for humans, interact with humans and so on.
At the end of the day the human form is what we build this world for and unless we want to rework everything to fit Daleks or R2D2, we have to build Cybermen and C3PO to fit into the world we already have.
Just saying that in Jetinsons authors imagined bot cleaners as robotic metal lady with eyes etc but actually we have small roomba circles driving around our homes. Isnt it what are you asking for?
They do both.
Sometimes industrial robots are very specialized. Ever seen car building robots?
But also though… In many cases, especially in places where robots and humans will be working in tandem… in the places everything is designed to be human friendly… human shaped. In those places it makes sense to have the bot be shaped like what the space was designed for.
Case in point. A brick shaped bot isn’t gonna be very good at doing the dishes. It can’t even reach the sink.
Sure, robot forklifts are very much a thing.
The attention is on more general-purpose robots designed to operate outside of a factory or warehouse setting.
Se have tons of robot-like robots on production lines, the human shaped robot idea is that it would be able to use any tool a human can to do a wide variety of tasks instead of being specialised on a single thing
The point of the “dog” like robots is that they are good for a variety of terrain. It’s true that a simple box is much easier to make work, and thats what most robots end up being. Ones that are shaped like humanoid and dogs are much more rare in the market.
As for why, as it turns out legs are a versatile means of transportation that can allow for a fairly complex amount of possibilities for locomotion. 4 legs is a good number to start with as it keeps the amount of moving parts down while also making the issue of balance much easier. As for humanoid robots, thats simply due to humans naturally being interested in a human shaped robot. Both for aesthetic reasons, but also for the idea of a robot that can move anywhere a human could move
everyone is missing the real reason. we don’t have in-home robots to help us because the guys that can make it happen have slaves that do all that work already – they are so out of touch that they can’t think of a useful product. they project and only come up with “robot buddies”.
It needs to be human shaped so it can take your job.
The simple answer is that it’s marketable. People like the human form and associate humanoid or animal-shaped robots as advanced and therefore desirable. Even though as you pointed out its more efficient to use different forms.
Its good to mention that many companies already make robots that aren’t humanoid and they’re highly widespread. Robot arms in factories, roombas, logistic robots in warehouses, etc.
The reality is that most of the robots we build aren’t humanoid. Only the ones you see in viral videos are, because that’s the type of stuff that goes viral.
We use robots in manufacturing (think cars), logistics (think automated warehouses, delivery robots), cleaning (think roomba), even warfare. None of those are humanoid or dog-like.
The only scenario where a humanoid, upright-walking robot makes sense is when they’re supposed to replace a human in an environment that is made for humans, and are meant to perform a huge variety of tasks.
Creating such robots presents the opportunity to refine and develop more versatile robots as the unique challenges it poses ultimately breed innovation.
But there’s more than that. Robots have been around for a while now and we know that robots designed for specific tasks in environments built for them can be very efficient and effective. But that comes at a cost, because you have to build specialised facilities for the robots to function. Humanoid robots are able to use existing infrastructure and equipment and work in spaces designed for humans, increasing their versatility.
Instead of having to build an entire space made to work with specialised robots you have humanoid robots that can be put anywhere and use human tools and equipment and get their job done.
The desire to build human-like robots stems from the fact that the world is, by and large, designed to be used by humans. Take stairs for example. Any robot with a regular set of wheels will struggle with stairs. There are designs that work around this, but they still fail on certain types of stairs — like a circular staircase. They also don’t do well over uneven terrain.
The bi-pedal humanoid form is capable of traversing a wide variety of terrain without issue. All of our doors and buildings are designed around the human shape and proportions. All of the every day objects we use are designed to fit in a human hand. Shelves and cupboards are accessible at human heights.
A robot with human-like attributes could navigate and take advantage of all this world has to offer. A robot-like robot is limited to an environment built to suit them. Both have value, but the more generalized form has more applications.
If you’re building a general purpose machine to operate in human spaces, it’s going to work best with a humanoid shape becuase we build our spaces around our body plan.
Special purpose robots tend to come in very non-human shapes because they’re optimized for one particular task.
As for dogs, the reason they give is pretty reasonable: they wanted something that could move across all sorts of different terrain and up surfaces that wheels or treads couldn’t handle.
Wheels are great for smooth surfaces. Treads work pretty darn well for lots of things, but they tend to damage delicate surfaces, they’re often loud, and they don’t handle tight spaces or stairs super great.
Legs work well for most land movement. They can’t match the speed of wheels, or the jumbled/muddy terrain stabilility of treads, but they’re the best all around, general purpose choice.
And, again, robots designed for specific tasks have locomotion optimized for that task, which often isn’t legs.
The big issue is general purpose vs special purpose. GP robots will tend to be humanoid or dogish, special purpose robots will tend to be all manner of non-humanoid non-dog shapes.
shareholders (morons) think that humanoid robots are more appealing – probably because focus groups (also morons) report this
humans love anthropomorphizing shit
bad scifi has put this idea in our brains
it’s all mixed together: FORM over FUNCTION baby
Amazon already makes box like robots that does tasks for it. The hype behind humanoid robots is that they in theory should be able to do everything a human can, not just one specific task
They build humanoid robots because the world they are supposed to function in is designed by and for humans. Its really that easy.