CARA 2.0 – “I Built a Better Robot Dog”
Comments
Animats
Tade0
> Motor overheating might be a problem. The dog, just standing, has its motors stalled under load, converting power to heat.
The Nao robot[0] had this exact problem and of course no way to fail gracefully. I recall checking its basic functions with my lab partner in college. I looked away for a moment and that was when it went down hard. Me and the other guy locked eyes in an "oh fuck" moment, as the robot was expensive and our thesis supervisor went through quite a lot of paperwork to have it funded. Fortunately it was intact and none of us mentioned this incident to anyone.
regularfry
It doesn't have to stall to stand still. Or squat, at least. With that leg layout it can safely rest against its backstops when the motors switch off. The drive motors, anyway. The hip motors probably still need to hold vertical balance, but that's intermittent, not a stall load.
JKCalhoun
"Temperature feedback would help if this thing has to operate for extended periods."
Rather than thermistors all over the place, perhaps an onboard program could calculate motor temperature by integrating current sent to each over time—assumed some degree of cooling (and perhaps here a single temperature sensor might measure ambient temperature of the environment… or could just assume "indoor temperature").
fennecfoxy
I would rather just have temperature sensors over the place. More reliable.
Animats
For a toy, you might just have a simple thermal fuse.[1] ($0.78 from DigiKey, plus "tariff may apply if shipping to the United States". About $0.15 on Alibaba, if you order enough.). For a working machine for field use, you want feedback to the controller, so it can slow down or rest or something when it overheats.
[1] https://www.digikey.com/en/products/detail/cantherm/RT121/28...
rmast
Train it like Disney’s Olaf where it can adjust how it walks/stands to keep motor temperature under control.
Gracana
That’s how industrial VFDs do motor overload protection, they just keep track of heat accumulation (based on current) vs dissipation (based on fan speed) and fault when a threshold is reached. Probably there’s more nuance to it, but that’s the gist of it.
barrenko
Could I pursuade you to expand on "Repurposed drone technology has done wonders for legged robots." ? Thanks!
cyanf
For legged, you want high torque and backdriveability (for shock absorption).
Agricultural drone motors like the eaglepower 8308 are ideal.
They’re cost effective, (~$80 from aliexpress) & you can pair them with a 3d printed cycloidal drive to fulfill both requirements.
Industry actuators are an order of magnitude more expensive than this.
Extra: If you go down this path, you’ll need a driver. The Xdrive is frequently recommended, but there’s a clone that’s significantly cheaper: https://makerbase3d.com/product/makerbase-xdrive-mini-high-p...
lastdong
This is pretty cool, thanks for sharing. I wish there was a (mostly) 3d printable Cara mini, but I’ll start with Cara.
regularfry
The layout is doable with hobby servos, but you'd need to patch in current sensing for that bit of the feedback. It's not terribly difficult conceptually but it's an extra complication that most servo power distribution boards don't give you.
You can also strap a capstan to the servo axle, if that's your thing. I've prototyped that myself in the past. You can go surprisingly far with an FDM printer, an SG90, and some dyneema bowstring. One thing I haven't tried is modding one for continuout rotation to get around the way the capstan drive limits the output angle you can achieve - I was happy reducing from ~180deg to ~45deg for what I was doing - but that's relatively well-trodden ground. Might pull that project out of the storage box it's languishing in at some point.
Animats
> The layout is doable with hobby servos
Mostly we're past that. Robotics with hobby servos sucks. Been there, done that, robot arm is in electronics recycling bin right now.
cyanf
Motor: https://www.aliexpress.us/item/3256804897857573.html?gateway...
The 90KV version is what you want.
JKCalhoun
"The 90KV version is what you want."
You mean the "KV90 color" of course, ha ha.
That we live in an age where "agricultural drone" is even a word pairing…
franciscop
Look at the original video/article, they used drone motors for the robot dog, by reusing the rotor/stator and rewinding the coils manually.
rmast
If you read the epilogue, they weren't able to achieve the under $1000 price goal. Total cost ended up being around $1,450. Pretty good price reduction compared to CARA 1.0 though.
Hypothetically if I were to want a quadrupedal robot to experiment with it's not an impulse buy/build, but getting closer to that point... whereas $3000+ is a hard pass (e.g. Apple Vision Pro territory).
regularfry
It's $1450 if you discount the construction time, as ever. Which ordinarily wouldn't be worth commenting on, but in this case it means rewinding 12 motors which just sounds like an exercise in tedium and hand pain.
15155
Only because they didn't know how to ask the vendor to do it for them.
I guarantee this vendor would be delighted to make them to spec at a 1ku volume, max. Rewinding isn't even a meaningful SKU distinction or line retool, it's a configuration parameter.
At 12 motors per product, it's easy to hit MOQ.
Animats
Rewinding drone motors for high torque and lower speed seems to be a popular thing. Here's the commercial process of machine winding motors like that.[1] That's a medium-volume machine, loaded and unloaded by hand, and adjustable for different wire and motor geometry. Here's a hobbyist version of a similar winder.[2] And another automatic hobbyist winder. Those wind under uniform tension, and with wire positioning, so you get a smooth, tight winding. This matters if you're going to use the motor much. Doesn't matter as much for short-lived toys.
There should soon be enough convergence that low cost, high torque, low speed models are off the shelf items. It's a great time to be building robots. Used to be all uphill on the parts side.
[1] https://www.youtube.com/watch?v=cMk2qFFcSho
drzaiusx11
There are inventor programs that'll literally ship you to Shenzhen to build connections to manufacturing sites and even provide you with a liaison, etc. I only know this because I was once in a program that did exactly this.
modeless
Mind linking some?
drzaiusx11
HAX Ventures (Formerly HAXLR8R) https://hax.co/
TroubleMaker (Conquer! program) https://troublemakershenzhen.com/member/apprentice-membershi...
Seeed Maker Camp https://www.seeedstudio.com/blog/2024/01/15/introducing-make...
I was an alumni of HAXLR8R before the rename.
dezgeg
Reminder that this was a student project.
KaiserPro
I love Aaed. Not only are the youtube video excellently produced and well researched, the website is a mine of information.
This is spectacular as a reference, which youtube isn't
pbmonster
The jumps are pretty impressive, this thing has some power. I'd be very curious how fast you could get this dog with some reinforcement learning for a proper transverse gallop gait - and if it converges towards a gallop naturally, or if it discovers some other fast gait patterns during learning.
Depending on the max speed of the motors/legs, giving it longer foot pads might be necessary for a good gallop. Intuitively, it looks a bit... "low gear" in the videos.
SamBam
I wonder how accurate a virtual model could be made of this, which could be iterated on millions of times faster.
My first project as a research assistant in AI was doing evolutionary algorithms on Khepera robots, which had a virtual Java implementation. We were able to evolve some pretty cool behaviors, although I don't know what would have happened if we had uploaded them into a physical Khepera robot.
leoc
As it so happens 'cara' https://en.wiktionary.org/wiki/cara#Irish is the Irish [Gaelic] word for 'friend' (probably related to words like Italian 'caro' etc.) so CARA is not a bad name for a robot dog!
erwincoumans
Very nice project. If you like this, the Pupper quadruped robot project by Stanford University is also interesting, with RL training pipeline as well and well-documented open source instructions for 3d printing etc. https://pupper-v3-documentation.readthedocs.io/en/latest/gui...
peterleiser
That is very cool. Their kit is $3,000 and the unaffiliated kit is $2,350, so definitely a commitment.
3form
People are all discussing the technical aspects of the device, which is great and all, but forgetting the one aspect I am in awe of: how googly eyes once again make everything 300% better.
defrost
A truism that sadly not all are onboard with: https://www.bbc.com/news/articles/c1wl5jp94eno
npodbielski
Seeing stuff like this, I am wondering why the hell I am doing software. This is much cooler than CRUD for DB data nth time.
horsawlarway
I loved robotics and participated in robots programs all through high school and college.
The answer is that jobs mostly don't exist for this kind of work.
The robotics jobs that do exist are basically large manufacturing robot development (think fixed place arms in car production), warehouse automation (amazon, etc), autonomous driving (originally agricultural, but now also things like waymo/tesla), and DoD style work.
And even then - the number of real positions is minuscule compared to other software roles.
So instead I build robots in my personal time as an expensive hobby (and man is it expensive...).
I'm somewhat jealous of the folks who've managed to monetize that expensive hobby by filming it and putting it on youtube, but I've also seen exactly how much time/effort/luck goes into being successful there - and I think it'd kill any joy I have for it.
voidUpdate
Hardware can be quite expensive and time-consuming, instead of just writing code (free) and running a command to deploy (quick) you have to get hold of good motors (expensive) and design and manufacture parts (slow)
npodbielski
You do not have to convince me! I already love it!
fennecfoxy
There's the mech/EE equivalent of crud, too.
Plus the demand is higher for software than it is for hardware most of the time. Pretty hard to find jobs in robotics compared to all the various kinds of software.
igleria
Same. I do have an excuse due to lack of space in my apartment, but once we move...
layer8
You could do non-CRUD software instead.
JKCalhoun
If you have any spare time…
tantalor
Rewinding the motors manually to increase torque is cheating if you can't scale that up and keep the cost down.
And, if you can scale that up, then why even mention it? It's not relevant.
proee
I'm sure he could contact a supplier that would change the windings for the same price if he went into production mode. So doing this for his prototype doesn't seem unreasonable. He's under a massive time crunch per his school semester to get this project out the door.
fennecfoxy
Yeah agree. Having to rewind the motors just adds yet another task to the list for someone trying to build this thing.
Also BLDC based motors in bots seem to be popular now, used to be steppers and harmonic drives from what I remember reading.
Idk why electrostatic actuators aren't getting more popular like HASEL and a few papers on ones using fiber pumps come out recently. I suppose the danger that the high voltages present hmmm.
But surely we can replicate muscle fiber actions magnetically, just need some sort of crazy micro manufacturing process to make the magnets.
fc417fc802
> surely we can replicate muscle fiber actions
Considering what muscle fiber looks like on the molecular level you're literally describing nanotechnology and the associated nanofabrication techniques. Surely it's possible to do but we don't seem to be there just yet.
abelsm
Awesome video. It's interesting, informative, and entertaining.
Founders I talk to that are doing hardware, broadly speaking, say it's a competitive advantage as it's not as crowded. Content like Aaed's will hopefully nudge more people into it.
JKCalhoun
Ha ha, that guy with his ever-underwhelmed wife. (Aren't we all this guy?)
elictronic
It’s because hardware has so many limitations that get glossed over vs. software. Manufacturing issues, sourcing issues, regulatory timelines, physical returns, inventory management, and oddities in the physical world are much easier to spot by an investor. Nikola motors and hills?
It can take 6 months once your product is finished to be legally allowed to sell it and even then it could be forced to be recalled or be redesigned. It’s a much more difficult environment to deal with vs I spent a few months throwing together this slop code, thanks for my multi-million dollar buyout.
RoxiHaidi
That's impressive! good job
Springtime
I like the wide layout of the site but just on a readability front on a widescreen monitor after the opening more narrow paragraphs it changes to full width text layout and those could benefit from a `columns: 2` in CSS to split them since reading long width paragraphs is a bit difficult.
crimsoneer
As a man who always desperately wanted a Sony AIBO (and is now only put off by the bonkers $300 a year AI subscription), I am unduly excited.
globalnode
its a fantastic achievement and the guy seems amazingly talented, is it a big achievement? leaves me feeling flat... maybe im just jealous lol.
dyauspitr
All I can think of is slippage and the lack of precision though I know I’m wrong because he spends a significant part of a previous video explaining why it isn’t an issue.
zzzeek
this guy is going to get a high paying job at some defense contractor somewhere and we'll never see anything he does again.
4gotunameagain
Oh you will see it eventually, but it will be the very last thing.
tamimio
Qugv looks cool and futuristic and has that wow factor, great in expos and milking investors’ money, but practically speaking? They are useless, they have the bad mobility of UGV (compared to UAV), and the bad endurance of UAV (compared to UGV), so they are the mix of worst of both worlds. There’s also other technical issues but yeah, only good for marketing, no matter how they try to push them in defense or whatever, they never materialize.
3 DOF per leg, so it needs 12 motors and controllers. Getting that under $1000 is nice.
Here's the US$18 motor: [1] Those things are getting really cheap. He did have to rewind it, though, for more turns with thinner wire. The manufacturer mentions that you can order with "custom Kv", which means you might be able to get a different winding from the factory if you order a reasonable quantity. Especially if you tell them that makes them "robot motors".
Motor overheating might be a problem. The dog, just standing, has its motors stalled under load, converting power to heat. Drones don't do that. Temperature feedback would help if this thing has to operate for extended periods. Remember yesterday's article on humanoid robots and their cooling problems.
The motor controller is nice too, and cheap at $49. Needed fixes to the firmware, but that's not surprising at the price. High performance motor controllers used to cost about $1000.
Repurposed drone technology has done wonders for legged robots. We're not quite at the point where limb drive hardware is off the shelf, but it's way better than it used to be.
[1] https://www.xntyi.com/tyi-5008-kv335/kv400-high-speed-brushl...