Looks like the Russians are pretty far behind in this race. Will the engineers get a Stalinist bullet to the head for this embarrassment?
Compare this with China’s progress. Interesting that the female-styled robot is attractively dressed considering that the one-child policy led to an imbalance of 10% more male children. The end-use is supposedly for commercial introductions which I guess is female gendered in Chinese culture. By contrast, the male robot is styled to be obviously mechanical.
In the U.S., Boston Dynamics now has an autonomous humanoid robot, the redesigned Atlas. This is apparently designed for strength-based applications.
Meanwhile, the Neo humanoid robot, which is designed to be a household helper, is being controlled by a live human behind the scenes. However, Neo does have good balance and fine motor coordination even though it isn’t autonomous yet.
I’ve never understood why we insist on creating “humanoid robots”. Do we think we are making surrogate children (subconsciously)? The quadruped robots are a lot faster, for example, and a lot more stable. They used those in Black Mirror, to terrifying effect.
Wheeled robots have told us most of what we know about Mars. (Pathfinder, Spirit, Opportunity, Curiosity, Perseverance, and even Sojourner).
A “domestic” robot wouldn’t need to look like Alice from the Brady Bunch. A wheeled chassis with optical sensors and multiple manipulators would make a lot more sense. I don’t expect my robot vacuum to look like a French maid; I expect it to perform its function and sweep my floors.
Exactly, but not forgetting that many many many tedious labors now performed by humans could be done with enormous efficiency by very wheely and or semi fixed platformish robots. Obviously highly repetitive industrial and agricultural work rarely need humanoid robots. But even within the human life environment there are many tasks probably best done by non-humanoids, e.g., weeding the garden, painting a wall, moving an obese semi-comatose person from bed to toilet to bidet to couch. I can imagine “intelligent” humanoid robots having a significant job in actually rerigging and jockying the non-humanoids around within human spaces to accomplish various tasks, starting with moving a roomba vacuum up and down stairs.
We are creating new systems for living with necessarily complicated “mechanical ecologies”, and healthy ecologies almost always have multiple species specialized in different ways as components.
@flyerboys these are tasks that require judgment and delicacy. The last one has the greatest need but also the highest risk of potential harm. I could see a “non-humanoid robot” in the sense of a pulley and railroad mounted on the ceiling with grasping arms below as long as there was human supervision. (Similar to rigs used to move heavy items in factories.) That would reduce the physical strength needed by the person. But I wouldn’t trust an unsupervised robot.
Precisely. The human form factor isn’t critical for pretty much any task. Dogs function in our environment quite well, for example. They are quadrupeds. Get a mechanical dog with optical sensors and a manipulator or two, and you’ve got yourself a stable robot that can move a lot faster than the bipedal ones. And less complicated (i.e. it’s a lot harder to balance a biped, so you have to have all sorts of complications to make that happen).
The real trick is the human hand. Perfect THAT, and your manipulator (on whatever platform) will be immensely more versatile.
Or go with the world of Fallout…this will work, too:
IMO, they are wasting a lot of time and resources trying to make them look like us when, functionally, it isn’t remotely necessary.
Exactly. And they are not focusing solely on humanoid form factor.
In my unboxed imagination, there will be many many form factors from bespoke and highly efficient to more generalized and “good enough”.
And, some robots will be “dumb” but VERY good at the desired task, while others will be ASI.
This YT hints at where I see robotics going.
Note he mentions Boston Scientific medical robotics, which includes robotic cardiac ablation for treatment of cardiac arrhythmias.
This is NOT likely to be a tiny humanoid form factor.
Brian also offers some ideas for potential investment.
He doesn’t mention TSLA.
I own TSLA stock in large part due to Optimus. As @1poorguy says the hand is important. TSLA is pursuing a functional hand which is absolutely required for SOME tasks.
For other tasks, suction cups are accessible.
As for hospital n nursing home environments, my unboxed imagination again sees multiple form factors.
The key will be sensors and AI guiding the hardware to accomplish a desired outcome.
Is there really a perfect way to fold laundry or do we simply do it in the easiest way our limited bodies allow?
Years ago, we washed clothes by hand in a manner that was most efficient for our bodies - because we had no other way. It certainly was not the easiest or the most efficient so we built machines to do it for us.
Why assume that the human form is the most efficient way to fold laundry? There is likely a more efficient form (there are industrial laundry folding machines already) than ours.
I will remind readers that AI vacuum cleaner robots can now go up and down stairs - without feet or legs.
But as we’ve discussed from time to time, there may not be much value in versatility. You only need a versatile body if: i) the robot’s “brain” is also very versatile; and ii) the robot will be doing multiple different jobs. Not at all clear that either of those things is going to be true, at least for a very long time and for most things that a humanoid robot might conceivably be used for.
This is not the assumption being made. If one were designing a robot for a single purpose, one would likely not think about the human form. But, we have offices, homes, factories, warehouses, etc., etc. which are designed for use by humans and which are likely to continue to be used by humans for some time to come. Hence when thinking about a multi-purpose robot to function in one of these spaces, it is natural to think of a humanoid form.
Note too that any deviation from a humanoid form creates a potential of not being usable in a human space. E.g., not allowing for climbing stairs or being able to climb rectangular, but not circular ones.
Yes, and we once had factories that were designed for use by humans - then we made multiple single purpose robots to replace us. Again, there seems to be this assumption that we should accept our own limitations in robot form instead of designing something that does the job better but with a singular purpose.
I didn’t say we perfected humans, I said we adapted the environment to humans. Now we are making robots that work well in that environment.
I neither assumed nor implied anything about folding laundry or any other activity. I said that humanoid robots are good in the environment where such imperfect humans exist.
I think it’s a prejudice (i.e. not the bad form of prejudice). We think we are the ultimate. Many people think we were designed by a deity, and therefore couldn’t be made better. Whatever the source of that conviction (design or evolution or whatever), we must be the best because we rule this planet. But human engineers design better, more capable, things all the time.
As another poster said, think a quadruped that has optical sensors (and maybe LIDAR), and manipulators as fine as the human hand. The hand is the key. With that, a single robot could go anywhere in your home, and accomplish any task you can. Probably more because you could make it stronger than a human.
Faster and more versatile, relentless, only limited by battery and ammo capacity. The Star Wars vision of bipedal robots (battle droids) in the precursor to the Clone Wars missed the mark. They will almost certainly more resemble dogs than humans. Or maybe spiders (6 legs, adds some versatility).
And with the head neck replaced by different folding levers (swiss army knife style) that can lift and pivot up from the body, some for sensing — vision, microphone, thermometers —, and others for grabbing/lifting lowering stuff, etc..
Yep. Amazon uses a gazillion robots that move around (along with the more traditional fixed-in-place robots). They do different things, but they share a common form factor. They’re all wheeled and low to the ground. They’re like hockey pucks or small wagons. That way they’re stable without having to mess around with something as complicated as bipedal balance.
Let’s say there are ten jobs in households and we buy a $100 specialized robot for each job. Total cost $1,000. What if a $500 humanoid robot could do all ten jobs but slowly. Total cost $500.
A new job arises, we get an over the air update to train the humanoid robot. Cost, let’s sat $10 vs. $100 for a specialized robot.
Think about the time dedicated to each job in a household. Most appliances sit idle most of the time.
Think about a smartPhone, it’s no longer specialized, it plays music, tells you the weather, and does a hundred different new things by just uploading an app.
Time for a mental paradigm shift from efficient specialized to multitasking anything.
