The LIGHTEST Robot Arm Actuator

[@dekutree64]

The other problem with bicycle cables is that even with the large bend radius, the friction rises significantly when they are bent. Skyentific did some good testing a while back, and as I said then, you could use a large servo to overcome the friction, but they'd probably wear out quickly since you're forcing them to move under tension. They work well for brakes because they stop moving as soon as the tension rises. Great idea using greased fishing line and PTFE tube. That should be extremely low friction, so it will probably last a long time. May need an encoder on the output side to compensate for stretching of the cord under high load though.

[@roTechnic]

Yeah, I didn't really notice the friction increase with this setup - probably because I wasn't putting a large load on it. I didn't think about them wearing out though - that's a good call. I think you're right about the encoder - I've got a feeling I'll need one on all the joints, not just the ones using fishing line - good job the AS5600s are cheap!

[@projecthivemind3239]

@@roTechnic Why did you rule out using DC motors with encoders? I am trying to make something similar, and that is what I was planned to use before I saw your video.

[@roTechnic]

@@projecthivemind3239 DC motors with encoders have the same drawback as stepper motors - weight. Although they weigh less than steppers, they still add weight to the actuator and will require some sort of reduction as they will not provide enough torque by themselves (unless they are overspecced and even heavier). The reduction (probably a belt drive or cycloidal drive as gears tend to exhibit the backlash I saw in the cheap geared stepper unit) will also add weight. Thinking about it, I might have misunderstood your question, if you just mean: why am I using steppers rather than DC motors then I don't have a good answer apart from that I am familiar with driving steppers and have never worked with a closed loop DC motor system. That sounds like a great idea for a future video though!

[@projecthivemind3239]

​@@roTechnic That makes sense. Thanks for your answer and sharing your techniques.

[@dekutree64]

@@roTechnic Capstan is another interesting zero-backlash mechanism, and would have good synergy with the bowden setup to get more torque without increasing the tension on the lines.

[@kiaranr]

Excellent content on a niche topic that is truly underrepresented on youtube. Minor quibble though: the background music is distracting

[@greendsnow]

exactly!

[@ianhoolihan2396]

Is there a difference between wobble and jitter,?

[@roTechnic]

Wobble is usually caused by beer, jitter by coffee :) But seriously, I tend to think of wobble as a low frequency movement, jitter as high frequency.

[@ianhoolihan2396]

@@roTechnic 😆😆😆 thanks haha I see now

[@ironman5034]

Hi, this is great, i just finished building a robot arm with 20kg futaba servos, i would like to try this drive system, can you add your cad files and a github link to your code if possible thanks

[@roTechnic]

You really don't want the cad and code for this - it was a quick, hacky experiment to find out if any of this was going to work. In my next video I'll be turning this concept into a proper wrist assembly and I'll be releasing the full cad and code for that on my github repo.

[@ironman5034]

@@roTechnic oh that's great, am looking forward to it!

[@m4t42]

Half a tennis ball 😂

[@roTechnic]

I thought you'd like that one :)

[@kaedkay2472]

Brother, this is truly an incredible series to watch. Very inspiring. I am currently working on my own cycloidal drives, and you are serving as inspiration and guidance at the same time. Keep up the incredible work! Have you thought about using a larger Nema 34 to make it more powerful?

[@roTechnic]

Thank you very much! I'm hoping that the NEMA 17 motors will be strong enough, especially with a bit of reduction. I'd really like to keep them all at this size as I'd love to create a larger version of the arm in the future by moving up to NEMA 34s and increasing the size of the other components to match :) But, there's always the option to upgrade them if my calculations are off :D

[@HornbyCrazy]

Been loving this series! I very much intend to build this arm. Really nice drive designs, compact and stable. Underrated for sure. Looking forward to seeing your work on the Inverse kinematics and code!

[@roTechnic]

Thank you very much! I'm currently debating whether to do the inverse kinematics by hand or use ROS and MoveIt. What do you reckon?

[@HornbyCrazy]

@@roTechnic I've not actually done either. I looked into Movit and it looks really impressive. You could have two series where you do basic programming first and then the more advanced coding later. More videos 🙌 I haven't seen many good videos showing Movit though so it would definitely attract attention of those out there who want to learn it. Might help your chanel get the attention it deserves, since your arm is one of the nicer designs out there for sure.

[@roTechnic]

@@HornbyCrazy Great idea! If I can figure out the manual inverse kinematics I'll do both!

[@slevinshafel9395]

i like how your voice sound. is warm and easly understandable. and nice content mecathronics.

[@chaosbreaker]

this series is so great, cant wait for t he next one

[@PotentStudios]

Close your eyes and it's like @reggiewatts is teaching you robotics.

[@klausnielsen1537]

I so wish you had time to work on this and show us your thoughts and progression. Oh well.... You are missed.😊

[@7alfatech860]

Neat idea to use fishing line. The bike cables are definitely overkill for the torque you need.

[@roTechnic]

Thank you!

[@martylawson1638]

Fine music wire can also work well for the core of control cables. Also allows a small amount of push. Might look at PEEK tube for the housing, it's a lot stiffer than PTFE. Fyi, you can use a PTFE tube inside a steel or aluminum tube to make "hard-line" control cable for any straight sections. Did this on my bike in key locations and it improved shifting a lot. Might also be able to use a single control cable if you fit constant force springs to both ends to hold tension on the cable.

[@roTechnic]

Oh, interesting. I'll see what I can do. Is the PEEK tube still flexible enough to bend round the other joints of the robot arm do you think?

[@martylawson1638]

@@roTechnic Make any material thin enough and it's flexible. (i.e. fiberglass cloth) I'd expect PEEK to be harder to bend than PTFE for the same size, but don't have a feel for how hard.

[@witoldkaptur59]

Great video as always :D Maybe the mechanism used in "remote direct extruders" could be useful here, especially with some closed loop controller to limit slop in the system.

[@roTechnic]

Thanks! I hadn't come across that mechanism before - I'll definitely be investigating it!

[@daliasprints9798]

Definitely check these out. I have some videos of mine in action. With proper gearing on both sides of the cable there's essentially no slop; mine operates with PA and retraction settings identical to a local direct drive.

[@roTechnic]

@@daliasprints9798 wow, fantastic. could you please let me know where you got them from?

[@daliasprints9798]

@@roTechnic It's the Flex3drive G5. The cable is specially engineered for this, so to get good results with a cheap off the shelf cable you probably need higher gear ratios. F3D does have the cable by itself available. The motor side gearbox is my own design.

[@roTechnic]

@@daliasprints9798 great, thank you very much - I'll order a couple and see what they are like!

[@DesignCell]

Found your channel today and it makes me want to start up a new project! I've had flexible shaft drill bit extension saved to an amazon list for some time with the idea of passing low torque to on end effector gear reduction reduction for dislocated drives. Very cool stuff!

[@MauricioHernandez-de8is]

What's up bro! Where are you? We need new cool videos, i hope you are doing ok

[@roTechnic]

Had a bit of a break, but I'm just printing the parts for the next video now! Thanks for your support!

[@emmanuelsheshi961]

nice

[@thejerrycheng]

Hi, thank u for the awesome videos! Is there any chance you could make a FOC motor video in the future video for an upgraded robotic arm? I have been really interested in making robot arms but hard to find any channel explained as clearly as yours

[@FilmFactry]

Your content is EXCELLENT. I will build this because I am fascinated with moving the heavy motor off the end effector.

[@normarainyahmad1367]

Hi, nice idea. Can i have the black pulley/spool 3d models file. The idea of that pulley so nice.

[@satyakambaruah]

Nice way to decouple the motors ... could we use a rotatory shaft too like the way we use in a Dremel tool.

[@connormarsh6339]

what are you using for the graphing of the motor target vs actual position

[@roTechnic]

I'm writing them out from the arduino on serial and then just using the serial plotter in the arduino IDE to graph them.

[@connormarsh6339]

@@roTechnic sick! I started working on my own 6axis arm (the goal to play chess and other board games) around the same time you started this series and ive learned so much from u!

[@sovahc]

Thank you, Engineer!

[@HOSTRASOKYRA]

Awesome! I am waiting for the next step!

[@ziad_jkhan]

I keep looking for subscribe button every time to then realize I'm already subscribed

[@roTechnic]

Hahaha! You could always create a new account and subscribe from that too!

[@richardurwin]

Great idea awesome!

[@dandil]

the reason you got poor results for the servos is because there's a control loop in them. you need to give it time to settle to the commanded position. there may even be overshoots and it's not great if there's load on the joint when static, but a good integral controller servo will eventually settle to commanded position. You can't just treat it like a stepper lol

[@roTechnic]

Interesting! So, would that work if I wanted the servo to be constantly in motion? like moving through a cycle of positions with smooth movement between them?

[@flinhadlee1944]

🄿🅁🄾🄼🄾🅂🄼 🙌

[@PhG1961]

Waw, impressive ! Well done, also very entertaining ;-)

[@roTechnic]

Thanks Philip! Glad you enjoyed it.

[@stefanguiton]

Exellent!

[@AltMarc]

Wouldn't a brushless gimbal work better?

[@roTechnic]

Maybe, but I couldn't find one that wasn't either way too expensive or too heavy. If you know of one, I'd love to see if it works for this application!

[@meltonermeltoner]

very nice!

[@meltonermeltoner]

where is the 3d models ? :)

[@roTechnic]

what you see in the video was really hacky and thrown together quickly to do these experiments. I'm designing a proper actuator using these principles and will release the CAD/STLs in the next video

[@meltonermeltoner]

@@roTechnic lovely ! top notch work!

[@williamhuang5329]

Hanzhen harmonic drive gear , strain wave reducer , robot joint , over 30 years experience

[@810mb3Rg]

Check out Paul Gould's channel!!

[@roTechnic]

Will do! Thank you.