Got to say I love your style of videos, they're not over the top, dramatic or sugar coated. Instead they're a really good and fun to watch build and project log with you sharing not only what you've done but how, why and where things go wrong, right and future tweaks or ideas. Really appreciate your videos.
Great project but you really should have some fuse protection in your power distribution, as one short or a failed odrive could cause a major burn-up with the current available from that battery. I'd suggest something like an automotive blade fusebox, with a separate fuse to each motor, or pair of motors.
Noo he has a crew.... 3d printing expert -James bruton 3d modeling expert - James bruton Electrical department -James bruton Mechanical designing - James bruton IT department -James bruton Camara man - James bruton Editor - James bruton Sales department - James bruton
Really?! Damn! Designing in CAD, 3D printing (which always need constant adjustments, filament replacement, upgrades and staying on top of the newest in tech and theory), making and editing videos, keeping up with the latest automata tech and computer programming, testing, coming up with more ideas, eat, sleep and do this every week! You are a machine!
I want to thank you for this series, i am working on a robot warm that is open source and can be made realtifly easily and uses wood construction to reduce the plastic impact of the project. I have hitten a brick wall but thanks to your video i gotten quite some nice ideas to continue the work.
I've been watching for years now and James must be 3D printing constantly. He really needs a laser cutter. I can't imagine what his productivity would like like if a 10 hour print goes to a 30 minute laser cut. It changed my whole prototyping game.
A handful of ideas for you: Print a module with a mechanism that retracts the D-Sub connector to allow removal of the main controller module without having to remove another cover to remove the D-sub connector manually. You could even implement some kind of cam/tumbler lock to require a key to move the mechanism to prevent tampering. Having a removal mechanism also makes it faster to install and replace the main module with an alternative such as maybe one that is Pi/ROS powered.
like a single battery, only a single concentrated power motor, also a single position encoder, then be sure of the variable transmissions, and/or just imu per joint
As has been mentioned I'm constantly impressed with your dedication to this project, thank you for making it so all I have to do is print one of these with special hinges and telescoping limbs so it has a backpack mode and I've got a portable printed personal exoskeleton
If by weird "some weird issues" you mean getting absolutely clocked in the face by your own robot, then yeah v2 had some weird issues, lol. Love your channel and your work is amazing, keep at it man
Awesome work! Has anyone managed to reduce the backlash in these reducers? I noticed quiet a bit when just printing the stls, but think you can tune the tolerances of the discs (inner holes are
One way to do it is tediously adjusting horizontal expansion and outer Wall inset. Print the outer wall first. Between these settings you can close the tolerances to the limits of your printers motion system. On my printers I've gotten as low as .04mm average tolerances. It's a pain to do, but once it's going correctly you'll be getting almost exactly the dimensions you're trying to print. So you can change the hole size via the horizontal layer expansion. Than scale the gear very sightly to close the outer tolerances.
I would really consider having that e-stop Open a contactor that kills power to the entire system. An e stop doesn't work for very well if the thing you're asking to stop by pushing it is the thing that's misbehaving to begin with.
Now that I think of it, electrons are basically hydraulic fluid. And batteries are pumps. Meanwhile encoders and circuits are... valves?... Whatever, I just think it's insanely cool analogy between electric and hydraulic robot dogs.
I love your craftmanship-engineering, really nice design, great work. One thing i want to see is a head peace for the dog, it would be better for it's publicity.
Gotta say it's amazing you get results this good out of those taz5 printers. We have 4 lulzbot at my schools lab. They are not my favorite printers. Impressive project, keep up the good work!
its great to see it all together.... i LOVE that spring in its step.. maybe it will have a bit of bounce when it walks and it'll look like its happy lol :)
Just an idea - you might consider running just one set of power leads from the battery terminal to the just one of the 3 the ODrives in a bank then daisy chain the power lead to the other two.
@2:45 how wickedly fast that thing can move when it sees an opportunity to get even... You should treat your hardware better (no torture tests) than maybe you wouldn't have to worry about bad karma!
steel power studs are fine, but might i suggest at least copper or brass nuts and washers for the intermediary hardware between the power in and power out to reduce resistance for less waste heat (in the studs) and better battery life. all copper hardware would be best of course but just replacing only the hardware that holds down the main wires on the bottom will still be beneficial.
+1 for more IK maths next time. The previous ones were good but I'm still not quite there with it (although probably need to do a rewatch). Mega in awe at the sheer complexity of the wiring on this one! Not sure how you are keeping track of it all!
This is amazing, you are Rapidly closing in on a truly disruptive open source robot platform. At this point the major weakness is electronics and wiring and that's a HUGE achievement and a fairly trivial problem at the social level. Just need some product niches filled, and even in this video with that pro unit it would appear that's going to greatly improve soon also. After you finish open dog maybe you could work on a 3d printed exoskeleton good enough to replace a wheelchair or a strength enhancer for physical therapy. IMO the most versatile form of robot other than a flying drone would be a wearable exoskeleton that doubles as a humanoid robot. Exciting stuff.
hi i've seen taht for a possible leg, i don't know how backlash would factor in but i think it could be programed to be quite springy, but know idea about the complexity of realisation. sorry for the poor writting. And thank a for your work and also the "open" part, really cool!!
Since driving the power transmission system has to take into account the mass (inertia) of the transmission components themselves, it will be interesting to see if the (relatively) heavyweight cycloidal drives you made will end up impacting 'dynamic' motion - like maintaining balance or recovering from perturbations. Also, have you thought of putting sensors in the feet to register pressure? "How much of my weight is on this foot?" Or even a binary "yes/no this foot is/isn't on the ground".
The motor controller is aware of the force at each joint. The springiness is all done in the Odrive. Getting that information out of the controller and using it won't be a high priority task until the robodog leaves flat floors behind and tries running on piles of rubble.
James, just wondering, how many 3D printers do you have, those parts are in the 3D printer perspective, huge, at least a 20 hour print for any of the large parts.
I watched all of your videos. I guess you didn't mention it or i missed it. But im wondering how much does it weight? And what is odrive current settings so how much power is required to run this beast?
Have you looked at the Ambidex actuator? As far as I know, it's out of a robotics lab in South Korea, and it's backdrivable. I'm sure you've got a big list of neat backdrivable actuators by now that are candidates for dog n+1 but I thought I'd add mine to the pile lol
For future Opendog iterations, rather than requiring backdrivability, would it be practical to use a higher gear ratio but with some kind of spring/compliance in the legs - this + creative use of IMU(s) would permit some kind of "virtual backdrivability"
I feel so bad for saying this, but I nearly spit out my drink in laughter when it clocked you in the face
You can't say that the legs aren't fast enough. 2:54
Spoiler alert.
"Sometimes there were still some weird issues"...
oof.
When the robot dogs want retribution for Boston Dynamics...
What did the 3-d printed ball-leg say to the face? SLAP
James: "... some weird issues"
OpenDog V2: "And I took that personally!" *SMACK*
th-cam.com/video/_mstbNizOjIy/w-d-xo.htmlyyu
Was that a slap or a punch tho
Got to say I love your style of videos, they're not over the top, dramatic or sugar coated. Instead they're a really good and fun to watch build and project log with you sharing not only what you've done but how, why and where things go wrong, right and future tweaks or ideas.
Really appreciate your videos.
He also admits his limitations- very down to earth intelligent guy. I want to be like James when I grow up ( :
thank god James isnt one of these screaming youtubers who follow only one rule: loud=fun :D
th-cam.com/video/_mstbNizOjIu/w-d-xo.htmluuu
Great project but you really should have some fuse protection in your power distribution, as one short or a failed odrive could cause a major burn-up with the current available from that battery. I'd suggest something like an automotive blade fusebox, with a separate fuse to each motor, or pair of motors.
That is true, probably one fuse per motor. It could be 60A though.
@@jamesbruton 8-10 bucks (USD) per fuse seems a lot cheaper than the motors/controllers :D
oh my god i didnt even realise that nothing here if fused!! man i stick fuses EVERYWHERE in my projects
@@ale6242 me too, and I'm just a woodworker!
th-cam.com/video/_mstbNizOjIy/w-d-xo.htmlytt
I like this dog as it uses passive dynamics in its walking.
Looks more natural than the boston dynamics in that yellow one.
th-cam.com/video/_mstbNizOjIj/w-d-xo.htmlhgf
Those electronics covers that fold out with all the wiring stored inside are really nice!
Odriverobotics, send this man some upgrades for the next robot dog, pls!
itd be cool but also as the more expensive the hardware gets, the less people will be able to recreate it
@@bbenny9033 True, but the other dogs are still there, if people want to get started.
th-cam.com/video/_mstbNizOjIj/w-d-xo.htmljhyy
2:55 robot uprising has begun :) anyway, beautiful engineering as always James, well done and very nobel of you for making it opensource
Do you do all this stuff alone or do you have a lab assistant crew helping you out? Cuz, man! Your workrate is INHUMAN!
It's just me!
Jesus Christ. RESPECT!
Noo he has a crew....
3d printing expert -James bruton
3d modeling expert - James bruton
Electrical department -James bruton
Mechanical designing - James bruton
IT department -James bruton
Camara man - James bruton
Editor - James bruton
Sales department - James bruton
@@jamesbruton damn! I hope you enjoy doing all this.
Really?! Damn!
Designing in CAD, 3D printing (which always need constant adjustments, filament replacement, upgrades and staying on top of the newest in tech and theory), making and editing videos, keeping up with the latest automata tech and computer programming, testing, coming up with more ideas, eat, sleep and do this every week!
You are a machine!
open dog v2 always reminds me of a bulldog and is quite cute in its movements, i can't wait to see how v3 turns out as it looks so promising so far!
V3 looks like more robot and less leg from the side, due to how tall the frame is, I think it looks good
I have keep watching 2:55 . I needed that laugh. Seriously funny. Thanks for leaving that in there.
It went into kill all humans mode. Jonny 5 alive
Seems more like the actions of a cat than a dog lol
That thang socked em
Behaves like a Robo Horse
2:54 did that robot dog just abuse its owner? That's too funny.
It's making up for all the testing at Boston Dynamics...
GIVE IT 5 YEARS, and I'll be riding the Open Dog 9 like a pony, cus this man is a genius!
I want to thank you for this series, i am working on a robot warm that is open source and can be made realtifly easily and uses wood construction to reduce the plastic impact of the project. I have hitten a brick wall but thanks to your video i gotten quite some nice ideas to continue the work.
2:53 That mules got some kick! thank heavens it was flexible filament on those feet.
Thanks!
I love how you 3d printed a work stand for this project.
this is a beautiful build. i’ve seen “professionally done” electronics in much worse shape with cabling and connectors not as well thought out. kudos!
th-cam.com/video/_mstbNizOjIj/w-d-xo.htmlut
I've been watching for years now and James must be 3D printing constantly. He really needs a laser cutter. I can't imagine what his productivity would like like if a 10 hour print goes to a 30 minute laser cut. It changed my whole prototyping game.
Prints can be made overnight and when he prints daytime, he can do other stuff in parallel. It is all about planning the workflow.
A handful of ideas for you:
Print a module with a mechanism that retracts the D-Sub connector to allow removal of the main controller module without having to remove another cover to remove the D-sub connector manually. You could even implement some kind of cam/tumbler lock to require a key to move the mechanism to prevent tampering. Having a removal mechanism also makes it faster to install and replace the main module with an alternative such as maybe one that is Pi/ROS powered.
the red and white look great, amazing work, both in quality and quantity 🤍
Amazing progress James, I hope you are ok after being hit in face by that naughty dog.
Looks much cleaner than the version 2 I love it! I’m really curious to see it working. You are making a great work
th-cam.com/video/_mstbNizOjIj/w-d-xo.htmluuy
like a single battery, only a single concentrated power motor, also a single position encoder, then be sure of the variable transmissions, and/or just imu per joint
Nice. James you have built everything but your own motor....
One of the best projects on TH-cam.
Well that a nice slap 2:55
Ooh dear, those unprotected screws for distributing the voltage made me nervous having shorted those kinds of batteries before....
The Odrive pro reference @7:30 ; should be on every mechatronic hobbyists' wishlist.
As has been mentioned I'm constantly impressed with your dedication to this project, thank you for making it so all I have to do is print one of these with special hinges and telescoping limbs so it has a backpack mode and I've got a portable printed personal exoskeleton
Looking better and better!
If by weird "some weird issues" you mean getting absolutely clocked in the face by your own robot, then yeah v2 had some weird issues, lol. Love your channel and your work is amazing, keep at it man
Awesome work! Has anyone managed to reduce the backlash in these reducers? I noticed quiet a bit when just printing the stls, but think you can tune the tolerances of the discs (inner holes are
yeah one option would be to print them with thighter tolerences and sand ore machine them to the corect tolerences
I'm pretty new to 3D printing, but you might be able to adjust the tolerances with horizontal expansion if it's really small gaps.
One way to do it is tediously adjusting horizontal expansion and outer Wall inset. Print the outer wall first. Between these settings you can close the tolerances to the limits of your printers motion system. On my printers I've gotten as low as .04mm average tolerances. It's a pain to do, but once it's going correctly you'll be getting almost exactly the dimensions you're trying to print. So you can change the hole size via the horizontal layer expansion. Than scale the gear very sightly to close the outer tolerances.
Have you calibrated your horizontal expansion?
th-cam.com/video/_mstbNizOjIi/w-d-xo.htmluyt
2:54 I had forgotten about this happening with Open Dog V2, now I get to laugh at it all over again!
I love the cleaner design and wiring!
I would really consider having that e-stop Open a contactor that kills power to the entire system. An e stop doesn't work for very well if the thing you're asking to stop by pushing it is the thing that's misbehaving to begin with.
Now that I think of it, electrons are basically hydraulic fluid. And batteries are pumps. Meanwhile encoders and circuits are... valves?...
Whatever, I just think it's insanely cool analogy between electric and hydraulic robot dogs.
3:56
Nice, lightweight, engineering there. Impressive.
I think Florian over at Voltlog might be working on CAN bus projects. I think some sort of interface for his vehicle.
I love your craftmanship-engineering, really nice design, great work.
One thing i want to see is a head peace for the dog, it would be better for it's publicity.
The productivity of this guy … 🤯
You're getting the hang of this! Very impressive build!
Gotta say it's amazing you get results this good out of those taz5 printers. We have 4 lulzbot at my schools lab. They are not my favorite printers.
Impressive project, keep up the good work!
th-cam.com/video/_mstbNizOjIj/w-d-xo.htmlhytrthj
10:39 Looks really steady. Congrats!
its great to see it all together.... i LOVE that spring in its step.. maybe it will have a bit of bounce when it walks and it'll look like its happy lol :)
2:54 the time where he got punched right in the forehead by his creation 🤣
This made me chuckle 😆
Just an idea - you might consider running just one set of power leads from the battery terminal to the just one of the 3 the ODrives in a bank then daisy chain the power lead to the other two.
2:54 - SkyNET activated lol
That looked painful though, glad to see you can laugh about it now xD
This is looking really excellent James!
2:54 LOL
I think he was sleepy and didn't wanna wake up xD
@2:45 how wickedly fast that thing can move when it sees an opportunity to get even... You should treat your hardware better (no torture tests) than maybe you wouldn't have to worry about bad karma!
steel power studs are fine, but might i suggest at least copper or brass nuts and washers for the intermediary hardware between the power in and power out to reduce resistance for less waste heat (in the studs) and better battery life. all copper hardware would be best of course but just replacing only the hardware that holds down the main wires on the bottom will still be beneficial.
2:54 Children are like that sometimes.
So intriguing. I was really not expecting when you were calibrating v2, and it punched you in the face. Yikes! Great work on your innovations!
Missed the ‘That’s all for now’ ! 😀
Very cool build! Looking forward to the rest of this series!
th-cam.com/video/_mstbNizOjIi/w-d-xo.htmlu66
+1 for more IK maths next time. The previous ones were good but I'm still not quite there with it (although probably need to do a rewatch). Mega in awe at the sheer complexity of the wiring on this one! Not sure how you are keeping track of it all!
That clip of the dog kicking you in the face….🤣🤣🤣
This is amazing, and I'm learning so much for designing my own body...
Amazing as always, James. I am really looking forward to seeing this design in action!
th-cam.com/video/_mstbNizOjIj/w-d-xo.htmlust
Leg locking blocks for servicing a live machine would be a good idea. could be very dangerous lol
Wow opendog v2 kicks hard
Brilliant ! - Great Video James - Love the setup of this one!
I can’t Imagine how much money just those 6 odrive motor controllers cost. That’s over $1000! Not even counting the motors!
Robots are expensive... makes me sad...
This is amazing, you are Rapidly closing in on a truly disruptive open source robot platform. At this point the major weakness is electronics and wiring and that's a HUGE achievement and a fairly trivial problem at the social level. Just need some product niches filled, and even in this video with that pro unit it would appear that's going to greatly improve soon also. After you finish open dog maybe you could work on a 3d printed exoskeleton good enough to replace a wheelchair or a strength enhancer for physical therapy. IMO the most versatile form of robot other than a flying drone would be a wearable exoskeleton that doubles as a humanoid robot. Exciting stuff.
What about mounting the legs at 80 degrees instead of 90?
Should help with side wobble.
Can’t wait for the next episode! This is looking great.
James, you don't really need the level converters for the LCD - 3.3V Teensy will drive a 5V display just fine as long as you don't read values back.
I like how the yellow robo doggie legit was like don’t touch me n punched you for a second I thought some ai was showing
That dog has a mean kick
James the kinda guy to say "holy moly" when he sees someone vaping
Your printers look so solid 👌
solid. better. good evolution.
hi i've seen taht for a possible leg, i don't know how backlash would factor in but i think it could be programed to be quite springy, but know idea about the complexity of realisation. sorry for the poor writting. And thank a for your work and also the "open" part, really cool!!
OMG! i believe spot from boston ! is 65K wonders what you could do with the cost of one !!!!! keep on have fun !!!
2:54 ''you're done, you're done.'' 😂
It's all fun and games until the cops replace their K9s with weaponized versions of this
ideally they save cop lives, and fulfill the same role with the same oversight, same risk to people being arrested.
the design of the endplate is calling out for googly eyes to cover over where the bearings sit :P
2:53 I did not see that before, ouch :D
Man, you look like you took that shot to the head like a champ. @3:00
James!! I never realized you use ubuntu. Love to see other linux users in the wild!
Dude took that blow to the face like a champ!
Since driving the power transmission system has to take into account the mass (inertia) of the transmission components themselves, it will be interesting to see if the (relatively) heavyweight cycloidal drives you made will end up impacting 'dynamic' motion - like maintaining balance or recovering from perturbations.
Also, have you thought of putting sensors in the feet to register pressure? "How much of my weight is on this foot?" Or even a binary "yes/no this foot is/isn't on the ground".
The motor controller is aware of the force at each joint. The springiness is all done in the Odrive. Getting that information out of the controller and using it won't be a high priority task until the robodog leaves flat floors behind and tries running on piles of rubble.
MAN! That is awesome!
Congrats on the million+ subs!
Nice robot dog
OpenDog V2 obviously a southpaw :)
Looking great 👍😀
James, just wondering, how many 3D printers do you have, those parts are in the 3D printer perspective, huge, at least a 20 hour print for any of the large parts.
Maybe not. Keep in mind most of these parts are not printed at 0.10 layer height / 0.4 nozzle.
2:53 dud gets railed hard and took it like a OG
I watched all of your videos. I guess you didn't mention it or i missed it. But im wondering how much does it weight? And what is odrive current settings so how much power is required to run this beast?
People in the 80s: I bet we'll have flying cars in the future
2021: 3d printed open source dog
Wouldn't it be ironic when in the end James just build this to have a springy chair that follows him around.
you are a good guy, nice vids too
Huh, wasn't expecting you to be an Ubuntu guy. Color me pleasantly surprised.
Have you looked at the Ambidex actuator? As far as I know, it's out of a robotics lab in South Korea, and it's backdrivable.
I'm sure you've got a big list of neat backdrivable actuators by now that are candidates for dog n+1 but I thought I'd add mine to the pile lol
2:55 open dog V2 : i hate you but i love you *smack*
a little disappointed you didnt hit the e-stop while it was standing. would be very interested to see what happens
Awesome progress!
For future Opendog iterations, rather than requiring backdrivability, would it be practical to use a higher gear ratio but with some kind of spring/compliance in the legs - this + creative use of IMU(s) would permit some kind of "virtual backdrivability"
he tried that in v1 but it is a lot easier to do it that way