68 comments

[ 4.7 ms ] story [ 21.0 ms ] thread
Bipedal drone. Its not the same to stabilize on the slackline with rotors than without.
It is the same. The propellers take the same role as gyros. It just uses atmosphere to create lift to become buoyant to make up for the fact that it is not very powerful. Once it is buoyant the interaction between legs and blades seem like the same type of stabilization as any other robot- fast micro adjustments of various potential-filled motors to stay on course
Not quite it’s much easier to balance on a slack line if you can put a hand or even just a few fingers on a stationary object. The propellers offer similar benefits.
> if you can put a hand or even just a few fingers on a stationary object. The propellers offer similar benefits.

As do gyros

(comment deleted)
Wow, what a ridiculously misleading headline. Caltech should be ashamed.

Thanks for clarifying things. I wonder if I could have a drone drag a string on a slackline and talk about my balancing robot.

Who cares how it's stabilized, it still has all the same advantages that a regular bipedal robot would have. Being pedantic about the mechanical "purity" of the design seems counterproductive.
It's not exactly being pedantic. If I saw a headline that talked about someone with a shattered spine learning to walk again, but it omitted the fact that the walking is accomplished with an exoskeleton, then it's pretty misleading. It can still be cool (I totally want an exoskeleton), but the headline is deeply inaccurate.

After all, the thing has "drone" in the basis of its name.

Not sure that bipedal motion is all that effective. Could have just put wheels instead for faster navigation and of course fly over terrain it can't navigate like its already doing.
In theory, flying and hopping (via flying) can handle all walking scenarios. So if a robot can fly, that's it.

In practice, I am guessing, walking or riding is far more power efficient.

Flying greatly reduces the weight carrying capacity, and is in theory more susceptible to high wind environments.

There’s a dozen “far future” use cases were a bipedal machine will be more acceptable to humans, including in home maid and artificial body.

This can be much heavier than something that had to actually leave the ground would be.
...and also fly over obstacles.

Leonardo is a drone with two legs.

Cool.

Nice.

The idea of having stabilisation near the top of the frame well above the CoG rather than down at the ankles or hips seems like one of those obvious things that is easy to miss.

I wonder whether/how that simplifies the rest of the control mechanisms.

I can't unsee the Grim Fandango pidgeon in it.
Cool, but impossible to tell how good it is without knowing how much of the weight the drone part is dealing with. It could be "Leonardo, the drone with legs that can touch the ground" but we can't really tell.
This could be great for package delivery.
It’s interesting we never see birds or insects doing this hybrid walking flying locomotion. Perhaps some species of dinos did it.

Update: An ask Jeeves search suggests ostriches use their wings for balance but I couldn’t find any videos. Can anyone find such a thing? Perhaps it’s misinformation.

North American quail do exactly this. They don't fly for long distances; maybe 50m max, or a hop out of danger or over an obstacle. Most of the time they walk, and they have ground nests.
Jungle fowl (from which we get chickens) can't fly long distances, they can only fly in bursts to escape predators, get to a roosting site etc.

A jungle is a good example for this, a larger, ground based animal that can boost jump (for all intents and purposes) to get out of trouble, but doesn't need to fly through dense foliage where it would be tricky for a larger animal.

> “It’s interesting we never see birds or insects doing this hybrid walking flying locomotion.”

it’s energetically too expensive for basically no benefit (in daily life), so that’s why it’s limited to escape situations where the mortal danger outweighs energetic cost.

this robot is also impractical beyond its research value for the same reason, too energetically costly. it’s too energetically costly because it’s constantly actively actuated, with two different sets of locomotors needing coordination on top of that, to achieve dynamic stability. animals rely on passive dynamic stability (akin to springs and dampers) to conserve energy and extend actuation time/efficiency (due to evolutionary pressure).

Hens, ducks, geese, pigeons all use their wings for various movements. City pigeons might evolve into penguin-like birds. Lazy bastards don't even get out of your way these days.
Chickens sure use their wings for short bursts of speed and even to jump over obstacles or onto trees etc. Scare a hen sufficiently and it will fly a few feet away. Like littlte caricature dragons though they're too plump and their wings too small to really fly. They kind of fly-run instead.
Chicken can fly but their thrust to weight ratio is terrible because want them to have more meat on their bones.
wild turkeys can really fly-jump high - for the night they fly up to the 30 feet high branches. Looks very impressive.

The robot in the OP is very inefficient for most of the practical tasks except where such mixed capabilities outweigh the inefficiencies, ie. like some military applications.

Roadrunner mostly walk and run, but can fly/glide when needed (going over an obstacle or escaping a predator).
Lots of room for optimization - hops are exaggerated height; props running even when on the ground.
What is this? A flying robot? No. To me, it looks more like a walking drone!
A high-heeled robot with a levitation spell.
What's after singularity?
Looks a bit like the main antagonist out of the Alien series...
That is absolutely terrifying. Can you imagine the Terminator movies where the Terminators can fly? I would have called it ridiculous before I saw this...
Robots are about as terrifying as a vacuum cleaner. They're fundamentally dumb machines that give an illusion of something more. Pull back the curtain and you'll see a potato brained CPU, ready to take instructions like a well trained dog.
Lots of people train dogs specifically to harm humans
Guns are even less intelligent than robots. They are still terrifying when someone points one at you.
The whole point of Terminators was that they could look and act like humans so they can infiltrate.

There were various flying war machines in Terminator movies. Notably the opening shot of the first Terminator movie shows a flying platform killing people with lasers, and in Terminator 3 there was an experimental drone in one of the labs in the facility where Robert Brewster worked.

Robocop comes to mind when looking at this thing
Me: "Leo, please bring me a cookie."

Leo hops off the table picks up a cookie and walks over. Unfortunately as it gets near the sofa one of its propellers makes contact with the cookie ...

Me, being showered in high speed crumbs: "Leo, please bring me another cookie"

Leo: "If you take a cookie I will be forced to add one full day to your sentence."
Me: "Leo, please bring me my printout"

Leo hops over the the printer and lifts the pile of newly printed pages. Unfortunately as it gets near to my desk, the paper is caught in the downdraft from Leos propellers ...

Me, manually sorting my scattered fifty page printout back into page number order: "Leo, please play the sad trombone noise"

Why not just fly?
Flying with props takes constant energy (unlike gliding). Standing takes very little. This is a research platform that’s somewhere in between.
Somewhere in between but very close to flying and quite far from walking.
That's not riding a skateboard. That is dragging a skateboard.
Agreed. This is not a bipedal robot. It's a drone that they added mostly useless legs to.

From the video, I'm not sure it could even walk at all without the propellers spinning.

It can't, they mentioned that.

Agree with your assessment that is a drone with useless legs.

(comment deleted)
This is a novel solution. Why should a robot replicate a human skeleton and muscle system. If the species could develop propeller mutations through evolutionary processes then perhaps we would have. I would say a drone with useless legs is an apt description for a bird.
Because it's a waste of energy. It's novel yet impractical.
Exactly, there is a reason birds don't have 20 lb legs!
Just curious, and this might sound naive but this is a serious question.

Are drone "blades" (not sure of the correct term) more efficient when they aren't enclosed? I understand this is a test unit and there are myriad safety precautions taken when testing. But my first thought when seeing open blades is what happens if it falls, trips or otherwise goes out of control? Then you have rapidly spinning blades going helter skelter to deal with. Not just hitting into any soft humans standing around. But the floor, tables, doors, wires, expensive computer equipment. It seems a bit daft unless it has a purpose that I've missed.

Most drone propellors are extremely lightweight, meaning that the damage they can inflict is very limited.
They have very high RPM though as my finger can attest when I mis-judged catching my DJI done once - the cut was surprisingly deep for light plastic props. I'd hate to see what a more substantial prop could do.
Generally, ducted propellers are more efficient.

And yes, these blades are effectively rotating knives. Even small drone motors and propellers are very dangerous as the motors have a lot of power and high RPM, and the more high-end ones are made of CFK and have sharp edges. As these four do not only lift a drone, but a whole robot frame, I'd assume they'd easily cut through a finger.

This looks completely useless. What's the point?
From the accompanying video:

"The goal is to give unprecedented walking ability and to solve problems posed by hybrid locomotion".

https://youtu.be/DhpMlI8jb5o?t=33

The idea that this is a bipedal robot is a joke. Makes you realize how ungrounded some of these research labs are - all hype, no substance.

This is a drone with a huge amount of weight for these "bipedal legs".

With things like this, it's worth keeping in mond that setting up a robot to do solo tricks is infinitely easier than getting a robot to interact with the real world in even the "easy" seeming ways that humans and animals use.
Something about this form factor that makes robotapocalypse seem more impending.
If robots can now ride skateboards and walk slacklines, I'd be interested to know what skaters and slackliners are planning to do now that they have suddenly been made useless.
I'm sure they will be able to reskill as robot service technicians
What is the point of walking if you can fly?