openDog V2 Part 2 - Building a test leg

[@jamesbruton]

Please please watch part 1: kzbin.info/www/bejne/pKjUhJ2DqdmWick and also my previous Robot Dog R&D: kzbin.info/aero/PLpwJoq86vov9Aogc6PseW8BUe2nQyZESZ Patrons and Channel Members have next week's video already (it's part 2 of balancing Strandbeest).

[@pearcomputers]

Gosh ur so cute :D

[@catseye10000]

next make a dog with wheels on its legs so it can turn in to a car

[@leerussell6536]

Call it Open Bulldog.

[@vbart1776]

Great to see you go a light asYou can so that you can eliminate nonlinearities in your models.

[@TOIMwet]

mister i'm working on a project and i realy intrested in you disign do you want it to put it on thingiverse because im just 13 years old and i chose tinkercad but its not good for my project if you want to mail my about it to ask what i want to do with it you can maby mail my [e-mail] thats my school mail and i hope you understand my inglish becasue im from holland/ the netherlands ps: your videos are realy good and i learn a lot of it

[@UnexpectedMaker]

I'm really enjoying these deeper dives into your design and thought process - And much clearer understanding of the evolution of the parts! Thanks :)

[@jnevercast]

I approve this message.

[@MKamTech]

cool

[@johnathanclayton2887]

I'm loving the aggressive optimisation! That's what I love about engineering. Keep it up!

[@generico366]

It’s worth considering that removing volume from a body in CAD can actually add mass to the final print, since a lot of the actual bulk of the part is in the perimeters of the part.

[@pushkarajsalunke3461]

Definitely, but if its planned properly you can avoid a large number of top layers, bottom layers and infill which are not contributing to the strength and stiffness. By careful planning one can introduce new walls to take the load in more directions then a infill or a top layer.

[@nathaniellangston5130]

Even though this robot will be more expensive than the previous R&D low-cost robot dogs. I am DEFINITELY much more likely to try and build it because it will still be WAY WAY WAY WAY WAY cheaper than a Boston Dynamics one and I won't have to be a celebrity to get one! I love this project!

[@callummacdermid1715]

This seems to be the most promising version so far! Love the iterative design you’ve taken over the lifecycle of OpenDog! Keep up the good work

[@rizalardiansyah4486]

Yes, I can see his dog designs are actually getting better from version to version.

[@MiniLuv-1984]

This is shaping up to be very promising.

[@sky173]

Outstanding. Glad to see the design going in this direction. Looking forward to more.

[@isthissiddh]

oh man i love the way you share your thought process.

[@CaptainShanko]

Definitely I recommend creating a cross-brace to support both pivots for the pulley and also for the idlers as well. It wouldn't need to be very heavy duty or thick to do the job.

[@jamesbruton]

Yep - already being done

[@joesubbiani]

OD V2 is looking super sleek! I cannot wait to see this all come together and when you get it working all these other dogs will definitely have payed off.

[@SmokyFrosty]

This was such a great dive into your design process. Loving how this project is coming together!

[@jamesbruton]

thanks!

[@Zalex612]

I'm much more impressed with the direction that James has taken this lately. The product in this video, though not refined, looks much more professional with a touch of art to it. Actually testing things out first and putting serious thought into weight are huge wins in my book. Keep up the good work!

[@FusionSource]

Hey James, great update on Opendog, great 3D prints, they look awesome.

[@old_gregam8274]

A generative design robot would look so cool

[@captainTubes]

Omg do the organic topological stress analysis. That looks INSANE O_O. Just wing it, it would be fun!

[@pietrom2642]

I approve! I love using it

[@liamburgo23]

He could always try message Tom for help

[@MKamTech]

yep, agree

[@OwenBuckingham11]

@@liamburgo23 Yeah I feel like James doesn't reach out to the community for help enough, seems to think he has to do it all himself most of the time. Contact Tom!

[@rizalardiansyah4486]

I don't think it is the best thing to do now. I prefer doing topology optimization as a last resort to reduce weight after the basic concept is finalized. James can still make greater improvements with trial and error since the concept (and the method of how the leg itself is going to work) hasn't been finalized. Edit: yes, i know he made a concept on how everything will work in first episode but it's still very rough concept.

[@anglerfish61]

your videos really got me back into doing some things like this myself. Very inspiring. thanks

[@josefroque5551]

the new design looks amazing

[@avejst]

Great design and good start Thanks for sharing 👍😁

[@joeytheghost4211]

I like the look of this, it has speed and power in a small package.

[@bernt6500]

Now it is evolving from a toy to a real robot ... like this videos!

[@riddellriddell]

Assuming each leg has a max force of 10 kg, when a leg is retracting it is only lifting its own mass (3kg?) leaving most of its force (7kg?) spare if you add a spring or rubber band that constantly tries to straighten the leg (with a force of 5 kg) the leg will still have excess power (2kg) to retract but will have that extra power (5kg) when trying to lift the body of the dog effectively adding a third more lifting power to each leg at no extra cost.

[@VeraTR909]

To save weight you could add 'speed holes' (racing term). just drill/print loads of big holes in any part that is overbuilt. This is a classic and easy way to save weight.

[@jamesbruton]

Yep it is already happening.

[@adamfermor2669]

Can't wait to see open dog v2

[@twitte0king]

Finally a slim down cad model!

[@jamesbruton]

There's still more coming

[@FaithfulMC]

Can't wait, looking good!

[@FilterYT]

Thanks James, awesome!

[@tres155]

loving the v2 looks like it will be much more functional keep up the good work !

[@Karlemilstorm]

Very nice. Love following along on this project. I have watched everything since episode 1

[@LionOfJudahRules1]

Brilliant James! your videos are so encouraging and informative! This is a most excellent virtual classroom!

[@pushkarajsalunke3461]

Hello, Instead of making the parts slimmer, you can try and get pockets around the design to reduce the mass. Also most parts can be reduced in thickness but you may have to put collar around the mounting.

[@xaytana]

While yes, this is a solution, I think part of the reason to use slimmer parts is to reduce bulk in terms of volume, not just bulk in terms of mass. Ideal construction would be based on stiff plates, pieces as thin as possible, but also as strong as possible; this is why you commonly see carbon fiber plates in a lot of robotics, especially drones. If you can reduce the thicknesses of components, and the distances of parts going through multiple sections, such as rods for hinges, you're also reducing the mass as a byproduct of reducing overall volume; as those heavier pieces, like the 20mm extrusion, do not have to be as long as they currently are. Look at the MIT's Mini Cheetah. Its actuators are composed of the motor, 10:1 planetary gearbox, and encoder in a singular housing. Both leg actuators are mounted coaxially, and mounted perpendicularly directly off the hip actuator. It's design could be made entirely out of plate material, though I believe some of their parts may be machined and/or printed, but plate construction is still entirely possible. This design is minimal volume and minimal mass, due to the assembly and construction of the robot. If any of the parts were bulkier, this would increase mass as the volume is increased, due to component sizes changing. Also look at Boston Dynamic's Spot Mini. It goes from MIT's Mini Cheetah's basic design, but rather than using metal, it uses plastics, rather than metals, for further mass reduction, based on the design of using shell pieces with a mostly solid lower leg. So while yes, you can just remove mass from a design, this would not be an optimal decision, as you also want to reduce volume to further reduce component mass. You want designs to be as efficient as possible in mass and volume, and a lot of this relies on efficient component placement and component sizes. A lot of the overall weight of this robot dog is going to end up coming from the body and hips, as it's _very_ inefficiently designed, which is a problem with all the dogs on this channel. Putting the hip actuators as close to the center of the body as possible would be optimal, which would also mean having left and right hip actuators side by side, and having front and rear actuators being coaxial with each other; just like how MIT and BD does it, as it's the most optimal design. From that design, the body could be lightened to be a very basic skeleton, rather than having extrusion running the full length of the dog. This would also put all of the core electronics within the center of the dog, and this could be designed as just a tray that also has perpendicular supports for the hip actuators. This would not only reduce a lot of mass, but it would also reduce a lot of unneeded volume, while also simplifying the dog, and concentrating it's mass center to the body's volumetric center. MIT's Mini Cheetah, and by extension BD's Spot Mini, and every dog following those designs, are the most optimal designs for a quadruped that mimics most mammalian leg structures. But this is where the downfall of a pulley system comes in, it's inherently more bulky than the kind of actuators other people are using, both in mass and volume. This also scales the more you use it, where the MIT/BD design concentrates the actuators into a very volume-efficient package, you can't do that with the pulley design, so it becomes very inefficient very quickly. This then adds more mass, as you have more components, and because the knee is used as part of the reduction, rather than being a 1:1 ratio, you're adding more mass and volume to the knee, which also affects the physics of the robot. This series on dogs is a good process of showing development and showing different thought experiments. But if you _really_ want a robot dog that can be the best it can possibly be, it's best to look at what companies and research groups have put through development, for the most optimal product possible. Just look at the change from MIT's Cheetah or Cheetah 2 to Mini Cheetah, there's a massive difference, and a lot of that difference was optimizations of how the robot works. We might eventually see something similar for openDog, as 2.0 is a step in the correct direction, though I wouldn't be surprised if we end up with a dog that looks like Mini Cheetah due to it's optimizations; also in part because there's clones of those actuators now coming from China, so building an _exact_ Mini Cheetah clone is more than possible, though we'd more than likely see a similar robot and not an exact copy. There's also variances in what could be the most optimal design, again look at Mini Cheetah and Spot Mini, specifically the lower legs. Cheetah uses a 1:1 pulley system to actuate the leg, while Spot uses a push-rod to actuate the leg. Is one better than the other? Who knows, I don't believe the two have been scientifically tested side-by-side; plus there's probably various use applications that would favor one over another. But both designs still concentrate their actuators within a volumetrically efficient area, at the hip joint. Other robots also do this, like OSU's/Agility Robotic's Cassie, which also uses push-rods to actuate various parts of its legs, while also finding other forms of compliance. It's best to borrow from what has already been proven to work.

[@apbosh1]

@@xaytana its 2025 I've just finished reading your comment!

[@pushkarajsalunke3461]

@@xaytana A very simple reason to not reduce thickness is avoiding unwanted play in perpendicular direction. When working with 3D printed parts its difficult to avoid the problem with smaller supports. Hence even though your solution sounds good on paper it won't be practical. If you want to keep it cheep and still make it work, some payoff has to be made.

[@pushkarajsalunke3461]

@@xaytana Ξανδρος Peaches One more thing I would like to point out is that mass is function of volume for printed parts. And reducing mass effectively does the same thing. But I am not satisfied with your comment on position of hip joint and use of Al extrusion. One has to keep design modular to allow trouble shooting and use standard components to keep cost to minimum. It is not possible to print parts with specific application and let it fail for some overlooked parameter and spend another day on printing (3D printing is painfully slow process, trust me I have made a business out of it). And spending money on manufacturing custom components to get little more efficiency out of design doesn't justify the DIY perspective of OC. To boil it down, Mr.James has done a very good job at keeping design easy to understand and reproduce.

[@johnh290]

Skeletonized parts would look so cool

[@mattiasfagerlund]

Looking very promising!

[@Pompiduskus]

awesome jjob mate !!!

[@geekyboi]

You should look a shelling the parts so they’re more like an I beam rather than just making them thinner

[@foldionepapyrus3441]

Not sure that will work - at least not easily with complex 3D prints. The layers mean the whole is not uniform in strength so the I beam profiles will with some layer orientations just want to explode, while in others you will get the expected results. Might be better off using actual Extruded plastic beams solvent welded or glued into 3d printed detail parts. Should be able to get all the benefits of 3d printed geometry and much of the strength of an injection moulded part. Will probably increase print time though as while each 3d printed part will be relatively small joiners there will be many of them and they will have to be high infill.

[@rizalardiansyah4486]

@@foldionepapyrus3441 I agree, it may even increase use of support material although it depends on printing direction.

[@Mitch3D]

You could CNC route carbon fiber plates (3-5mm) it would solve a lot of rigidity and weight issues.

[@MrTrilbe]

I was thinking fluting or fuller-ing the printed parts to save weight and keep most of the strength, but carbon fibre plates and rods would be a lot better

[@jamesbruton]

There will be Carbon fibre tubes in it, but the rest is 3D printed until it gets some way towards working - there's also ways room for openDog V2.1 etc

[@forcefeedback7839]

About reducing the weight: For the bearings, maybe Igus bushings will do the trick

[@3dmechanicaldesign]

Great work James! 👍

[@eldaine]

nice work SKYNET

[@zedfalcon6972]

9:30 christ, that thing's speedy

[@benntrowbridge]

Just out of curiosity, it looks like you only soldered the header pins that you planned to use on the rotary encoder (7:22). Was that just an example for the video before you finished soldering or did you only connect like 1/3 of the pins? Was this just a way to save solder and time at the expense of the solder strength? Why not use a smaller header pin or trim off the unused pins?

[@jamesbruton]

Yes it was just for testing, I'll need other pins eventually.

[@benntrowbridge]

@@jamesbruton Ahh, that makes sense. I hadn't seen it before, so I thought I should ask- just in case anyone else was curious. Thanks for the quick reply and the great videos!

[@OM222O]

Great video series! Keep up the good work! I also had some suggestions that you may find useful. First you can try breaking the reduction into multiple stages.Best thing would be making a planatary gear box that's attached directly to the motor and have some teeth on the housing for the belt. Then you can have all of your reduction in the gear box and just a 1:1 reduction with the belt that goes to the knee. This also allows you to reduce the size of the gears that the belt interfaces with, reducing more weight, as well as allowing you to change the gear ratio with minimal re printing of parts. To save more weight you can use a truss structure rather than solid plates. This holds most of the strenght at a fraction of the weight. Plastic is also not the best strenght to weight ratio material, I remember you had a CNC, if you do the same truss structure with aluminum you will have a lot stronger parts at the same weight or lighter parts at the same strenght.

[@jamesbruton]

I discussed this in part 1, I would probably strip the gears on planetary gearboxes if they were 3D printed at least. It's also heavier and more complicated. I'm trying to keep is simple and light.

[@OM222O]

@@jamesbruton I have actually done some simulations and found a small gearbox (8cm diameter, 4cm thick) that holds up about 15KG. This would be lighter than your current setup (I estimated a lot of things, especially the infill to be 50%, so I'm not entirely sure). Another youtuber called geardown for what has some pretty neat designs and demos where he shows even lifting an anvil! You won't need something that large and beafy but it proves the point that 3D printed gears can be quite strong. Another tip would be using better materials such as petg as PLA is very brittile. I'd also be happy to share my simulations and designs if they would help.

[@duret-robertlouis2973]

16:22 wait, isn't there an issue with the pulley of the hip joint ? It should be facing outward, opposite to the motor, right ?

[@jamesbruton]

There's more than enough rotation - the back has to be cut off on the outside to let the shoulder rotate past it.

[@duret-robertlouis2973]

@@jamesbruton oh alright, just the top and bottom teeth are enough. Thanks !

[@scream221]

Then there is no need for teeth on the joint at all. Only a few would engage anyways and you could just clamp an open belt at the top and button instead. I assume the teeth add some weight to the print as the outline is vastly elongated.

[@duret-robertlouis2973]

@@scream221 yes, or even simpler : fix the belt some other way (clamp, screw) on the side facing away from the motor. That way, no need for teeth at all

[@daOlliruo]

i would really like to hear how many hours you put into these builds. Your pushing out all these videos, and i'm always wondering how many fricking time your need to put into all the designs. i really appreciate your videos they are always really well made and refined!

[@jamesbruton]

Well it's full time basically.

[@Robothut]

Because the weight will be higher and the motors can run so much faster than you need I would changing the gear ratio to twice or 3 times what you have. Then the weight of the complete dog will not be a factor. Great job. Love the project video's.

[@jamesbruton]

I may need another 2:1 ration in the knee, otherwise it's tricky to fit in

[@markstevensfpv3722]

I started commenting to suggest topology, then just before I hit enter you were already discussing it 😂

[@ukrainewarroom]

since you last posted about this I have been doing a bit of research and seriously want to design and build my own. Looking at an HS tool head for my TAZ so I can print Carbon filament and other stronger filaments.

[@wktodd]

@James Bruton Just a thought James .You can double the gear ratio of those joints by using a pulley system - instead of a fixing loop belt to the leg/arm , use a length of belting that loops around two Small pulleys fixed to the leg , the ends of the belting can be looped bay to fulcrums near the motor. In fact by looping the end back to the leg you'd get a three fold increase in leverage

[@Innomen]

Excellent.

[@gistnoesis9116]

For my parameteric pulleys I use openscad thing:16627

[@World_Theory]

I think the value of what can be done with a little bit of tougher material to reinforce a high value connection point is amazing. And don't forget that there is a middle ground between a bushing and a ball-bearing; a roller-bearing. (I don't know if you have any in stock, though.)

[@dynorat12]

looks great a little bulky but a great test peace thanks James

[@jamesbruton]

Yes more refining is coming as mentioned.

[@StevenIngram]

I can't wait to see how this goes. And the open dog project has been a really interesting look at the evolutionary nature of rapid prototyping. I mean, how many designs have you gone through in the last year? 4? 5? :) I wish I knew how to design parts and 3D print, I might try something like this myself. Though on a much smaller scale. LOL Little tiny robot dogs. :) Open Chihuahua. LOL

[@taleg1]

Would aluminium plates have the needed strength? It would be cool to see a comparison and learn more about the plus and minus of using either. How thin can the metal be and still be stronger than the 3D print. There is one other thing that is prickling my mind about that leg and it feels like an idea on how to both strengthening and loosing weight the whole thing, but the idea isn't fully formed yet or so deep in my mind that I can't access it. Nice work though.

[@akselwilliamdanenbarger7969]

Thanks for showing fast how you are drawing in the cading program.

[@HoRiOnS]

Good luck with making everything as light as possible. You can also consider using a rectilinear infill pattern inside many parts as a filler that will keep it structurally solid and strong while getting rid of a lot of weight. You're almost there man, I can feel it. :-D

[@HoRiOnS]

Check out this conversation about infill patterns and shells: kzbin.info/www/bejne/faishJqud52ZhK8 Very interesting in regards to making your robot as light as possible.

[@jamesbruton]

Yes it is

[@dtec30]

Can you use a Teflon runner like an engine timing chain guide?

[@BebenX]

Very exciting!

[@LSUtiger607]

Something to consider to solve the weight issues. Lay out on a large flat table a carbon fiber sandwich with thin foam in between. Then use fusion 360 to print out 1:1 templates to cut out the parts you need. You could also carve and shape blocks of foam and wrap in carbon fiber for more complex shapes. You could also buy pre made carbon fiber rods at the hobby store for axles. Very strong and a fraction of the weight of steel bolts.

[@tiporari]

Just a thought for the future, you can print a jacket that fits the outside of the outrunner motor which has a pulley around the circumference. Separating the rotor from the stator for safety you can drill and tap screw holes in the rotor and screw your pulley sleeve to keep it from rotating. Would be more compact, and increase the pulley diameter making it easier to get it to grip the belt.

[@rodrigomartinez3937]

This is amazing pulleys are a great choose but big one's dont need to be full size, you can make only the operative angle and integrate it into the keg part

[@michaelmorris6699]

I think IGUS make plastic bearings, roller and plain ones, you could maybe use those instead of steel skateboard bearings. If you can get the bolts to be shorter then you might be able to reduce the diameter of them too. Composite plates would be a lot lighter than 3d prints for the same strength, but milling them would be an effort, might be able to get them done professionally for cheap but i'm not sure

[@NitishYadav-lb7zc]

Hey what about collabing with skyentific💥

[@jamesbruton]

This isn't a small servo driven robot, the dynamics are entirely different as detailed in part 1

[@SnorwayFlake]

@@jamesbruton I think his skill set would be greatly advantageous for your drive system. Adding planetary gearboxes in the large pulleys would be right up his ally. Also most of what him and you are doing are servo driven systems, he also seams to have a great mastery of IK and jakobians, the first you have implemented using a geometric approach, the latter would be greatly beneficial in doing joint force control for more dynamic movements. Also I think it would be wise to have the back hip joints all the way in the back, instead of towards the center, its just an educated guess, but I think it'll make for an easier time getting the CG in between the two legs when the dog is walking.

[@kaiinagaki8119]

I really agree with that!

[@NitishYadav-lb7zc]

He is also doing vision stuff and deep learning stuff

[@SnorwayFlake]

@@NitishYadav-lb7zc Ohh yeah, but apparently this is a big boy servo system, so no need for that or those skills... all joking and sarcasm aside I think all those skills would be super useful :)

[@nelsonphillips]

you could use a much smaller motor for the hips with a high gear ratio.... speed will never be a factor with the hips

[@maloriezastrow971]

You should really have a copy of Machinery's Handbook. That would help you learn more about designing gears and pulleys and the forces they can take and all that.

[@__bennie_]

If you want to remove weight from the bearings, you can use www.igus.com/info/plain-bearings-iglide-product-finder as replacements, these are also made from plastic but wear resistant and can help make a better performance while reducing weight

[@coulombicdistortion1814]

I'm almost certain you can replace the 2020 shafts with hollow square tubing of identical dimensions for significant weight savings. In fact, it looks like you were already planning on using some similar extrusion for the second revision of the body.

[@jamesbruton]

Yep, it's just a test for now, I'll be using Carbon fibre tubes eventually.

[@nophead]

Yes extrusions are not great for resisting twisting due to being basically only connected by a tiny tube down the middle.

[@GafferTapeIsSticky]

Great work James! With the belt tensioning idlers, do you need 2 per belt? You might be able to drop some weight by going to a single idler.

[@jamesbruton]

For the knee I need two to make the belt fit inside the leg. When my structure is more rigid I may need none for the other axis.

[@captainTubes]

Love your stuff man 💪

[@gregwood911]

If I recall correctly htd belts might be patented, so you can really find design info on them. You can however download 3d models from sdp-si

[@mayuralakshan9130]

ඔක්කොම බලල තියෙන්නෙ.... පට්ට වැඩේ ඈ.... ත්‍රීඩී ප්‍රින්ට් එකක් ගත්තම..... හදලා බලන්නං 🥰😂

[@habibg2138]

Awesome

[@crabson1864]

This version of opendog looks like it is made of two parts front and back. It would be cool if you could put them on a some sort of tube (like a spine) that would allow front and back to rotate independently to some degree.

[@captainTubes]

Carbon box tubes instead of aluminum? CNC milled carbon plates with carbon standoffs and machine screws for rigidity?

[@jamesbruton]

There will e round carbon fibre tubes in it.

[@captainTubes]

@@jamesbruton Thank you for sharing ☺️

[@morilamorila2825]

is there a posibility to measure torqe dynamicly? like a "rotary load cell"

[@tiagotiagot]

Could you use the belt tension to measure the forces being applied to the legs?

[@jake_a_g]

Me pretending I know what's going on --> 🧐

[@Collin326C]

have you thought about using your cnc router to make carbon fiber parts instead of plastic?

[@jamesbruton]

Yes but there's always time for upgrades rather than cutting up expensive material straight away.

[@MichelSatoer]

Maybe you can also integrate counterweight -springs in the joints. That probably will relieve a lot of stress on the motors. And have you tested the strength of a skinny printed part versus a bulky part but with way less infill (to get about the same weight)? Or maybe with real hollow cavities inside the parts. It might be stronger with the same weight.

[@DerSolinski]

Looks good, I think this version will have the mechanics sorted out. But you really should brace the motors with additional bearings, potential shock forces will wreak havoc if you don't. I know it is super annoying to do so, but repairing over and over again is more so.

[@jamesbruton]

Yep, that's is happening, Patrons have details already.

[@nightcrawlersgamingchannel5135]

Lol what time is it there cause it’s currently 1:30 in the morning and it said it came out only 44 minutes ago

[@jamesbruton]

It's always 1.30 in the morning for someone.

[@RX--yr6zm]

great❗ I want to make my own robot. What kind of book should I study?

[@StuFrankish]

Ever wanted to do a video with Boston Dynamics on how they approached certain challenges or how they achieve dynamic stability? Just seems like a logical thing to do or are you more interested in the discovery & learning process and deliberately not getting involved with Boston Dynamics and Spot?

[@jamesbruton]

I'm not sure if they'd have me ;-) If I were going to try something like that then I'd prefer to get something that at least walks reasonably well first also.

[@wesleymays1931]

I have a question about the encoder boards: If removing and shorting SMD components was an option (required, even) the whole time, why did they even have a jumper header in the first place?

[@scream221]

How do those magnetic encoders work? I first assumed the magnet will me mounted off axis, but that would only give one interrupt per revolution. With magnet and Sensor on the axis I've no clue 🧐

[@csteinbeck]

Very nice progress. Seem like this robot should be able to jump off the ground for the first time (which I like most about the Cheetah :)). And: You might have mentioned it somewhere, but what material are you printing? Would there, some fine day in the future, when the whole design is finished, be some high-performance material that is lighter but as durable? Or maybe saving material with that generative design feature is the way to go ...

[@jamesbruton]

I'm just using PLA for now, but Autodesk have offered to help me out with industrial printing at some point.

[@Lucian_Andries]

Will you be able to make the legs and body as slim as the ones from Boston Dynamics and MIT? Or at least close enough, with what you have? I really want to see how those work...

[@TrudgeRC]

Hi James - not so sure about the maths at 15:58. If it's on 2 legs, and it weighs 15kg, that's 7.5 kg each right?

[@jamesbruton]

The load test was on top of the leg on the ground. So I'm talking about the forces on top of two legs not including the legs themselves.

[@bpark10001]

Where do you get the pulley tooth profiles for RTD belts?

[@jakejeffery2576]

With the number of robots he has made you could make ‘five nights at James Bruton’

[@gunalx]

tip there is an auto generator in fusion under scripts. or you could add in a addon that makes it easier to generate pulley systems.

[@StefanRink]

Did you try using cubic or quarter cubit infill? I find it brings way more strength to the prints so I end up using a lot less infill and making the prints lighter, yet stronger. Keep up the good work!

[@plemli]

Even these large diameter low kv brushless motors do not enjoy delivering static torque and may eventually overheat. The solution is smaller higher speed motors and large reductions, really, which is why harmonic drives are a thing while we wait for artificial linear muscles.

[@jamesbruton]

MIT Cheetah is a 10:1 reduction.

[@plemli]

Interesting, that's a fairly low gearing.

[@AJB2K3]

drill out the centers of bolts and threaded rods. there is a small quantity you can remove to reduce weight without sacrificing strength. Also, have you looked at the increasing shells and reducing infill trick to reduce plastic use and weight?

[@jamesbruton]

I just won't be using them, or using smaller ones where I can

[@alecwhardy]

Is there any reason you didn't use a 9235-100 motor. They are slightly larger but can offer double the power.

[@sonofzingo7]

Why do you use extrusion as shafts rather than using something like hex shaft?

[@rodolfoespino727]

Can you please tell me where I should find these motors?

[@RichardRamos]

Which magnetic encoder did you use

[@jamesbruton]

The AMS5047

[@ShipwreckedMonki]

Nice looking prints, Where do you recommend buying filament from James?

[@MattFowlerBTR]

He's got an affiliate link for Matterhackers in the video description 😁

[@ShipwreckedMonki]

@@MattFowlerBTR oooooft. International shipping on top of already expensive filament equals too much. Back to searching.

[@globoyquan5414]

Is there a reason for the legs being short or do you plan on making them longer? And do you plan on adding any type of machine learning or neural networks later on?

[@kremsnita446]

A robot idea! A robot that plays a video game!?