Cycloidal, Capstan, Quadrupeds, Arms - Robot updates 2021

[@irkedoff]

I want to see and play with all the new toys you made.

[@tituscassiusseverus6303]

It's lovely when you kids take an interest in learning your hobbies, but I wish they would buy their own damn Toys 🤣

[@PaulGouldRobotics]

Toys now days are boring.

[@HimanshuGhadigaonkar]

@@PaulGouldRobotics that’s true.. the new toys don’t make you think of how to use the toy creatively.. they are already moving themselves..

[@anotherwingover]

These are very interesting and novel designs! I am keen to learn more about these. It looks like the teddy way has quite a small bldc motor driver board. Which board did you use there?

[@PaulGouldRobotics]

The motor is Gartt ML5010 and the controllers are custom running Arduino on ESP-32 WROOM, FET Drivers and magnetic encoders. They are communicating with ESP-NOW. It is the board in the middle at 4:32.

[@dhruvbose8294]

Seriously awesome! What do you do on the sides to fund all this though?

[@PaulGouldRobotics]

Robotics is just hobby I do after the kids are in bed. The printers run 24hours sometimes. I reuse a lot of my stuff. I've probably spent $2000 over the past 10years. Plus $20 a month on filament. The hackaday prize finalist money helped too. I design lasers and high speed digital cameras for work.

[@dhruvbose8294]

@@PaulGouldRobotics interesting! Keep it going dude.

[@stevemarchando4132]

What an underrated channel. Gj as always

[@robertoguerra5375]

Nice :) i am most interested in the leg that has 2 servos in the hip

[@OhHeyTrevorFlowers]

I love to see the quadrapeds and novel gear designs.

[@PaulGouldRobotics]

Thanks

[@PhG1961]

Great overview and what an amazing achievement ! I prefer the robotic arm, with the 3-fase brushless motors. They have a lot of torque/rpm's and quick reaction. I'm looking forward to see your next big thing... ;-)

[@PaulGouldRobotics]

The robot arm is using stepper motors. The quadruped leg is using brushless. The NEMA17 and the ML5010 brushless weight the same. The brushless has 20x more power. Sadly I have no "next big thing", just making progress on these projects.

[@PhG1961]

@@PaulGouldRobotics Yes indeed, I saw the whole overview. I'm planning a robotarm myself, I'll go for the brushless version because of the more torque/power. Your design and prototypes inspire me. I would appreciate if you could continue making improvements on this.

[@PaulGouldRobotics]

@@PhG1961 I’ve got an update to the wrist coming soon. I’m not sure what to do with the end effector. Elbow is ready to print. No idea what to do with the two base joints. Will have to watch some Skyentific videos for inspiration.

[@PhG1961]

@@PaulGouldRobotics Cool ! Indeed, Skyentific is also one of my hero's !

[@socaldevin]

More capstan stuff, please!

[@PaulGouldRobotics]

Check out my Instagram for more capstan designs.

[@H34...]

Curious about your prosthetic elbow (and that exoskeleton you off handedly mentioned!) why would you want it non backdrivable? For something being constantly handled by and interacted with by a person, you'd want the compliance backdrivability provides, no?

[@PaulGouldRobotics]

Prosthetic elbow should not fall down when the power comes off. It could be switchable with a clutch or brake. Exoskeleton should be backdriveable with the power off as to not effect the wearer. Exoskeletons are difficult as they could break someone's leg if they go wrong. The forces are also high.

[@H34...]

@@PaulGouldRobotics To me a prosthetic and exoskeleton are very similar applications, and I would argue the prosthetic should be back drivable for the same reasons as an exoskeleton. Someone may want to manipulate the prosthetic while the power is off (maybe battery dies), or just so that it can bend and yield if it is hit, or maybe during a fall, rather than forcing the wearer's arm around.

[@besenyeim]

For the record: I'm a fan of capstan/cable drives. Cycloidal drives are great, but have disadvantages, especially when 3D printed. Have you considered trying out the "orbital drive"?

[@alexeymaslov3865]

What are disadvantages of cycloidal drive?

[@PaulGouldRobotics]

Efficiency is low. High backlash if the tolerances aren't perfect. Uses a lot of bearings. Slightly heavy.

[@besenyeim]

@@alexeymaslov3865 Needs tight tolerances, or it will bind up or have larger backlash. AFAIK, no method for backlash compensation. The basic design seems simple and elegant, but it is unbalanced, can't be used at high speed. A double gear construction is usually the solution, but that doubles the part count and complexity.

[@PaulGouldRobotics]

@@besenyeim I have a antibacklash design in an old video. If you use three disks it can be statically and dynamically balanced. It’s a little complicated but nothing crazy. Bearing count is high.

[@Robin-pq1cx]

@@PaulGouldRobotics I've enjoyed following your experiments for some time now. The bearing counts don't need to be high. iirc your cycloids have bearings on the roller pins - these shouldn't be required as the disk should be rolling over the pins. In industrial cycloid drives, the drive pins often have bushings rather than bearing. In my own 3D printed drives I just use 3 or 4mm pins that are pressfit into the drive disk. I only use two bearings, one for the eccentric on input and the other on the output part of the drive disk. I"ve load tested a 16:1 cycloid drive in the same formfactor of a Nema17 driven by a mediocre nema17 motor to lift 1kg at 300mm before the nema17 started losing steps. The biggest issue is cycloids really need precision components though there are ways to approximate that precision with 3D printed parts - drilling out holes to fit precision stainless steel 3mm pins. There is a paper by a Romanian(?) professor that goes into some detail on how what parts of the cycloid path bear the most and suggests that trimming the outermost parts of the lobes. Just trimming the outermost parts of the lobes reduces possible interference from low precision 3D printing enough to make these drives feasible. I will try to dig up the reference to the paper. found the link: imt.uoradea.ro/auo.fmte/files-2007/MECATRONICA_files/Anamaria_Dascalescu_1.pdf

[@piconano]

Sub'd +1

[@makersengineering530]

👍

[@manuelsousa9355]

Great work! I would like to see more of the slim cycloidal with BLDC ;)

[@PaulGouldRobotics]

There will be an update on this on Instagram in a few days.

[@dsg123456789]

I'd love to hear more about the series elastic actuator leg design and also the tiny capstan drive legs.

[@PaulGouldRobotics]

My Instagram has lots of new tiny capstan drive legs. Videos coming soon.

[@hybrayhem]

Mini servo leg looks interesting.

[@PaulGouldRobotics]

I'm working on that now. I have all the parts to finish it. Just have to finalize the hip joint and print 3 more.

[@SchulteMixKanal]

Like to see it too!

[@Fredz66]

@@PaulGouldRobotics Which model of servo is it ?

[@anonymouscoward9459]

Lots of nice designs , what one is your favorite ?

[@PaulGouldRobotics]

That's like asking which kid is my favorite but the brushless cycloidal quad is my fav.