3d printed quadruped robot driven by hobby servos, li-Ion batteries, a teensy and gyroscope. Remote controlled via bluetooth. (new code can handle carpet)
This is the most impressive servo-driven quadruped I've seen. Its movement is quite close to Boston Dynamics' dogs. And it looks like it could be very agile with the right control algorithms.
WOW!! this is awesome, this is absolutely the best homemade robot ever seen! seems very well mechanicaly designed. Automation on that just blows my mind.
smart putting the servos in towards the shoulders as much as possible as this lowers the moment of inertia and thus strains the motors less and probably makes stepping quite a bit quicker
Sadly not. Dont get me wrong, its awesome, but what you see here is hard leg moovement, the original is adapting every moove on the fly to what its sensors see. Thats the actually difficult challenge. Just making the legs moove so it odes basic walking is pretty easy.
@@harrysarso Im not telling, this thing is a bad walker, just the fact that it looks like its walking like the original is ver decieving. Its not like someone just did homself what boston dynamics needed years for. The difference is like one between a modern car and a motorized sateboard. Its really nice when you can build a board, much skill involved, but its still not a car just because it can moove and has four wheels ^^
great job, for some dampening and become more flexible mechanics, you would put a strong spring throughout rods of servos which manipulate the elbows it will bear some load with reduced servo stress.
Great job! I'm currently reengineering the legs of mine in a way similar to your robot. The closer the motors to the axes of rotation, the faster they can be moved since they have less inertia.
The issue I see with that is that the lower the motor is on the leg the harder it is to turn since the weight is lower. From my understanding you can counteract the inertia using a PID algorithm
Great job!!! It moves like living thing. I guess the author used accelerometer, hiroscop and software Kalman filter to generate control signals for servodrivers of legs.
Three letters OMG it looks like the boston dynamiks one. Very impressive!!! I saw down in the comments that a lot of people were talking about James Bruton. His robot dogs are not as impressive and I dont think they work properly. I think he should focus more on the speed of his robots. This is why they keep falling over and over. There are olso heavy which is olso a factor that causes the robot to be unable to lift its own weight. Excuse me for my english. Its because I am not a native speaker. I am from Greece
Very nice! I'd pay for the STLs (or better, CAD files) and code for that. You have it far enough along to be a great base for a coding adventure. I reckon you could get it cantering.
Which algorithm are you using for the gait control? And what corrections on stability maths are you doing from data of the IMU? It seems it is working so well for now, waiting new updates!
well @jamesbruton is trying to teach at the same time so everything becomes bigger and more exploded but i do agree somewhat that this guy has achieved something very cool indeed
@Martin Triendl It looks like you have a Trotting Gait. In a high level, can you explain how you implemented the Gait? Do the diagonal limbs mimoc each other? How did you figure out your step reach and speeds? Also is your balancing done by adjusting the shoulders regardless of the speed of the Gait? Or, is it more coupled with the Gait? Have you looked at using BNO055 IMU for your absolute orientation? It has the sensor fusion built in it. Amazing Job with this robot! I am super impressed.
Its a fixed trotting gate where the timing for stances and step is always same atm. The trajectory for each leg using same equations but is calculated independently and corrected in any process cycle due to IMU and Bluetooth input. There are two balancing routines, one for stance and one for step phase. Its controlled by the IMU. I tryed the BNO055 as well, but it has a drift and its to slow for balancing.
@@martintriendl4453 Could you explain a little bit the IMU-driven gait correction? I'm working on a similar quadruped and have implemented a gait planner that uses the raibert heuristic (put the foot forward x=1/2 * T_stance * velocity) without IMU, but it has trouble when the side to side oscillations build up. Thanks!
@@NathanKau Before you start any controlling... is the body in balance? Take a scale and check if you have same forces on all feet in standing position...
Hello what kind of pins do you use for the smaller joint connections, not the one ones you used for bearings. I’m not quite sure what they area called or what to look for. Thanks you!
Thank you. The problem with friction is not that easy. With the current code the servos are tormented too much when I increase the friction. I am currently working on a solution (suspension, force sensors, new code...).
Wonderful! An awesome job for sure. Can I ask, where I might find the STLs and code to follow your path to success? I would be neat to build one around the size of a house cat.
i think with these kinds of quadruped hobby builds the main issue is people focus on the ability to walk, some effort should be put into NOT walking... the constant stepping gate of these are kinda inefficient and are usually what causes the instability. a more manageable slower more careful ambulatory program might make it more stable and not look like an excited puppy. dont get me wrong though, this thing is amazing work and by far the best I've seen. I'm just saying it might be time someone looks into a slower paced more measured walking system rather than "stomp stomp stomp" at least thats my view.
This is the most impressive servo-driven quadruped I've seen. Its movement is quite close to Boston Dynamics' dogs. And it looks like it could be very agile with the right control algorithms.
OMFG...what a cute little thing!!!
Will you make the files avaliable for other people to build it? Please??? ^^
That's looking really fantastic! I can't wait to see more, keep up the great work!
WOW!! this is awesome, this is absolutely the best homemade robot ever seen!
seems very well mechanicaly designed. Automation on that just blows my mind.
I have to agree with the other comments, it's incredible work for hobby's servo ans single person work !!!! Congrats!
smart putting the servos in towards the shoulders as much as possible as this lowers the moment of inertia and thus strains the motors less and probably makes stepping quite a bit quicker
this is so much better than the robot dog/mini robot dog series! Thanks youtube for finally giving me a good recommendation.
As for single person work vs billion dollars company - it's excellet work. It looks and behave almost identical like dog mini by Boston dynamic.
yeah, I have to agree with you there.
"Dog mini", lmao
Sadly not. Dont get me wrong, its awesome, but what you see here is hard leg moovement, the original is adapting every moove on the fly to what its sensors see. Thats the actually difficult challenge. Just making the legs moove so it odes basic walking is pretty easy.
@@NIOC630 fortunatly this can be mostly if not entirly compensated for with good mechanical leg design
@@harrysarso Im not telling, this thing is a bad walker, just the fact that it looks like its walking like the original is ver decieving. Its not like someone just did homself what boston dynamics needed years for. The difference is like one between a modern car and a motorized sateboard. Its really nice when you can build a board, much skill involved, but its still not a car just because it can moove and has four wheels ^^
Nice work Martin!
Awesome work! As I understand, this is only a beginning. But so impressive beginning!
props m8, great work. Keep it up!
what a cute lil thing, it looks like it's excited to be there
I would like to know more about this. Good job !
Wow damn, that's really cool. I think the youtube algorithm caught your video :d
Sick bicycle by the way :D
That's incredible! Juat amazing.
Cool robot, and good handeling do it have to.👍🏻
Incredible work.
Looks great!
great job,
for some dampening and become more flexible mechanics, you would put a strong spring throughout rods of servos which manipulate the elbows
it will bear some load with reduced servo stress.
It's so cute. Good job bro 😎👍
This is pretty amazing
Nice video!
Martin is a genius in reverse engineering guys..we should applause and respect.
amazing work!!!
Great job! I'm currently reengineering the legs of mine in a way similar to your robot. The closer the motors to the axes of rotation, the faster they can be moved since they have less inertia.
The issue I see with that is that the lower the motor is on the leg the harder it is to turn since the weight is lower. From my understanding you can counteract the inertia using a PID algorithm
Nice job!
Boston Dynamics Called: They want Spot's puppy back!
IMPRESSIVE WORK!
Good work !!
Can you open code?
Great job!!! It moves like living thing. I guess the author used accelerometer, hiroscop and software Kalman filter to generate control signals for servodrivers of legs.
Who’s needs a to go out when you can stay in quarantine with this little guy
Ain that a cute little thing
I can't tell if the marching in place movement is part of the walking algorithm or an imitation of the BD's robots. Amazing work.
Wow your controls have really improved with the intro
watched it a second time - just as amazing to see it scramble around, puts alot of big project to shame
Three letters OMG it looks like the boston dynamiks one. Very impressive!!! I saw down in the comments that a lot of people were talking about James Bruton. His robot dogs are not as impressive and I dont think they work properly. I think he should focus more on the speed of his robots. This is why they keep falling over and over. There are olso heavy which is olso a factor that causes the robot to be unable to lift its own weight. Excuse me for my english. Its because I am not a native speaker. I am from Greece
Hyper little bugger! Good work!
Great job
Nice, keep it up. Thank you for sharing :-)
Very impressive!
Cool robot!
Very nice! I'd pay for the STLs (or better, CAD files) and code for that. You have it far enough along to be a great base for a coding adventure. I reckon you could get it cantering.
Supper robot broo . Fantastic
I have to say that the dance matches the music very well.
Nice and cool music
Subbed to see more of your content
Cool..Very... Great Work...
perfect, great job...
This is epic!
wow that is super awesome. i want to build something like that
Genius idea
Nice! Did you sand those parts? Were they SLA?
Impressionante!!!
Thank you and thank UA-cam for good suggestions! More such content, maybe how it was done too.
Very good 👍
Está relindo, te quedó fenomenal con un exoesqueleto de Aluminio sería más que resistente
good work
Outstanding..
Cute puppy, what kind of microcontroller did you use to program? PIC, dsPIC, Arduino, Picaxe or Raspberry π?
Small dogs are cute. Small robots are just as cute. :D
Man these seems like a miniature version of Boston dynamics..
Would love to build one.. can I have the 3d images file? For printing the parts
Thats great man......
cool 80's music
This is amazing work, congrats!
Any plans on open sourcing it?
pretty cool
Awesome!
Which algorithm are you using for the gait control? And what corrections on stability maths are you doing from data of the IMU? It seems it is working so well for now, waiting new updates!
It all custom code. The IMU works with a gyroscope and an acceleration sensor combined by complementary filters.
So cool, love this-- please give him (her?) googley eyes
Awesome
Hi Martin. How are you making the robot sense the edge/borders and then change direction
@James Bruton - look, it´s finished :D
this looks better and walks better than both of james bruton's open dogs.
@@cluberic You mean the 3 we've seen so far? He said the minidog rebuild can walk, but we won't see it until Tuesday, unless you become a patron.
well @jamesbruton is trying to teach at the same time so everything becomes bigger and more exploded but i do agree somewhat that this guy has achieved something very cool indeed
Being smaller i think it' needs a simpler gyro setup but Is a very well built project
@@MagivaIT you can teach once you fix the mistakes...
Amazing
He visto varios de este tipo, pero la terminación de este me gusta mucho
Omg that is so damn cool!
Это идеальное воплощение одомашненого псоРобота!
Тазобедренный сустав компенсированный!
Отличный подход. Код проекта нужно изучить
that is awesome!!! Are you using a custom motherboard or something like Arduino or raspberry based?
@Martin Triendl It looks like you have a Trotting Gait. In a high level, can you explain how you implemented the Gait? Do the diagonal limbs mimoc each other? How did you figure out your step reach and speeds? Also is your balancing done by adjusting the shoulders regardless of the speed of the Gait? Or, is it more coupled with the Gait?
Have you looked at using BNO055 IMU for your absolute orientation? It has the sensor fusion built in it.
Amazing Job with this robot! I am super impressed.
Its a fixed trotting gate where the timing for stances and step is always same atm. The trajectory for each leg using same equations but is calculated independently and corrected in any process cycle due to IMU and Bluetooth input.
There are two balancing routines, one for stance and one for step phase. Its controlled by the IMU.
I tryed the BNO055 as well, but it has a drift and its to slow for balancing.
@@martintriendl4453 Could you explain a little bit the IMU-driven gait correction? I'm working on a similar quadruped and have implemented a gait planner that uses the raibert heuristic (put the foot forward x=1/2 * T_stance * velocity) without IMU, but it has trouble when the side to side oscillations build up. Thanks!
@@NathanKau Before you start any controlling... is the body in balance? Take a scale and check if you have same forces on all feet in standing position...
what servos are those. they seem to be pretty fast and strong
Little big dog)))) Good! Very nice))))
Hello what kind of pins do you use for the smaller joint connections, not the one ones you used for bearings. I’m not quite sure what they area called or what to look for. Thanks you!
There are ball-bearings in any joint (six in each leg).
All other parts are 3d-printed.
Hey martin congrats on the project, can you share the project description along with parts please......
Which servo did you use? I was thinking to use 35 or 45 kgcm servo and additional encoder. What do you think ?
heyy thats pretty good
Boston Dynamics eat yer hearts out, great work, thanks for sharing :)
"Mom, can I have B o s t o n D y n a m i c s?"
"We have BD at home."
BD at home:
Jokes aside: really nice job!
That's looks amazing ! why you don't increase friction under the leg and what if we use low torque servo ?
Thank you. The problem with friction is not that easy. With the current code the servos are tormented too much when I increase the friction. I am currently working on a solution (suspension, force sensors, new code...).
By the way it's amazing 👌
Wonderful! An awesome job for sure. Can I ask, where I might find the STLs and code to follow your path to success? I would be neat to build one around the size of a house cat.
Me too where are STL files
0:10 it can beat Mayweather by them shoulder movements
amazing...
Okay. Do you want robot overlords? Because this is how you get robot overlords.
Wow it's like spot mini from boston dinamycs
what servos are you using?
What servo motors you have used?
I wonder what more can it do with Boston dynamics algorithm, gj💗
Nice work Martin. Do you mind sharing the 3D model files?
awsome..
could you send example of your code? do you use hardcoded algorythm or neurone network?
New battlefield 2042 trailer looks good
i think with these kinds of quadruped hobby builds the main issue is people focus on the ability to walk, some effort should be put into NOT walking... the constant stepping gate of these are kinda inefficient and are usually what causes the instability. a more manageable slower more careful ambulatory program might make it more stable and not look like an excited puppy.
dont get me wrong though, this thing is amazing work and by far the best I've seen. I'm just saying it might be time someone looks into a slower paced more measured walking system rather than "stomp stomp stomp" at least thats my view.
Which program do you use for your robots?