In order to finance my RU-vid channel and my projects, I am starting the consultation in robotics. So if you would like to ask me questions about my builds (or about your projects) you can do this through my website ( www.skyentific.com/book-online ). This service is not cheap, because I would like to limit it (with lower prices I would have to do this all day long, and I would not have time for projects/RU-vid). And of course, I cannot guarantee that I will be able to help you during this consultation, but I will do my best.
The biggest drawback of the “quaternion joint” is that it’s not actually kinematically defined with true rigid bodies. It relies on tiny amounts of joint slop and elastic deformation to act as a joint.
One of the chief drawback from mechanical point of view is that at specific angle in 3d coordinate, there will be excessive torque needed for delta movement. Similar to knuckle joint strains at 45°
If it were constrained by an extra armature, it seems it could be used as a constant velocity joint. But again it may be limited by torque loading... However I wonder if that could be worked from the other direction by using a CV-joint design that handles torque well as a robotic armature if it's not constrained like it would be on a vehicle drive line?
@@keixsy The design allows angles with lots of stress against specific joints. I.E. too many hard materials to flex and distribute the stress accross the entire structure.
I'd love to see how you worked out the inverse kinematics for this thing. That would be a great video on its own. These spherical parallel manipulators are just so mesmerizing to watch. They remind me of a delta 3d printer in some ways. Beauty and engineering. Well done. I think it'd be cool to stick a 360 lidar on the top of one and use it for slam mapping in 3d.
Thank you so much for depicting your wonderful project. Your work and effort are oustanding. Your movies have been providing excitement and inspiration for so many years. Thank you again!
Dear Dynamixel, hope you are reading this. I am already a fan of your products, using them for years now. It is great to see you support one of my favorite channels. Would be happier if you support this channel even more!!
This is awesome! It’s like a mechanical elephant trunk. It’d be awesome to see 10 of these joined together getting progressively smaller by a factor of 1.618. I’d love to see this in 6061 aluminum with the proper power and servos. This arm would be extremely helpful in space on the ISS or on a moon base. Imagine it on an automated gantry that can move in x y and z. It could put together entire habitats and move parts etc.
Lol, I spent an entire weekend in Fusion 360 recently "inventing" this very same basic joint structure. It goes without saying that there is nothing new under the sun and brighter minds have usually tread the same ground. Keep up the great work and thanks for sharing!!
@@dm5rkt Thanks! I just uploaded the concept I worked up. It moves in 3DOF, but would require linear actuation/artificial muscles to control. Not sure how practical. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-oOC0FxmM4WA.html . If you have seen my concept worked out further somewhere, let me know. I am not trying to be original, lol. Just want a compact solution for a "shoulder" joint. Of course my joint suffers from multiple positional limitations.
I watched a video about rolling joints and didn't even learn how to roll a single one. Kidding, this is a really cool mechanism. Thanks for putting this video together! :D
There is another benefit to this design, modularity. Your first 2 axis have the same configuration and the same hardware as the second 2 axis and so on.
Fascinating point you made that because you're using three motors for two degrees of motion you can use the servos to eliminate backlash. Also like how this design lends itself to easy cabling. Very cool. Thanks for sharing.
Just an idea, without undetstanding the geometrical operation in detail ... Couple the three axis with gears and drive two of them and let the third free spinning. Backlash elimination might bite the grass but you could use worm screws for the axis for better torque... Maybe the third axis needs a brake for holding
It's an interesting concept. Idk if you've seen what naver labs with the ambidex used this joint for, but very similar application. If you decide to expand on this project you might try changing the drive ratio from 2:1 to 4:1. That would increase your avaliable torque but at the cost of speed. Great work as always. Keep it up!
@@juliusfucik4011 yes very true. It's the design and CAD skills that matter. I do everything in open source OPENSCAD as I refuse to commit my designs to rent software.
Use a seperate servo motor drive for better current distribution. Here voltage is not the problem, current is , current plays important role in providing torque.
have you tried to replace the gears between the plates with worm gears to make it more stiff? The worm gear can move the upper gear but the upper gear can't move the worm.
Hi, i found new way to build reductor with planetar gear and ratio 354:1. May be it will be interesting for you ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-mkzmRiBHxcY.html
I would love to see this configuration done in aluminium and with greatly increased gear ratios. It feels like it could be 1:100 instead of 1:2, and still have usable top speed.
That mechanism would also be great for the spine of a robot, giving it a lot of mobility. Also, I wonder if this mechanism would be better if it is driven by cables rather than gears.
Beautiful and magnificent. I can't believe I've never seen anything beyond the literal rolling joint - with the cables/strings and also only with one axis. These multi-axis rolling joints are awesome. Great stuff!
I would wonder about worm drive from continuous rotation servo motor ,bit harder to code , but better ratio and no backlash. Can't figure out if a compression spring rather than the carbon rod would be better - maybe on the outside? Give it that old Tobor look, .
Instantly subscribed. Love this! Maybe you could use high speed DC motors with a very low gearing. Then for positioning you could use other concepts; perhaps include orientation sensors or some other kind of feedback.
Hi, I have been following your content and it has been very interesting and informative. Just a suggestion for this design- at every stage you just need 2 actuators instead of 3 as according to my understanding any 2 of the arms are active and the 3rd is kinematically coupled. It will reduce the weight of the stages and might help the robot dynamics.
In theory you only need two motors per joint, in practical application you need three for holding strength. There's also reasons why tentacle-like arms do not approach their design like this, but rather use air bladders or wire actuators that have a larger motor in the base, with a more simplistic design that uses a flexible non-compressible spine between the actuation discs; not to mention higher arm rigidity and less complexity in these models. Though the alternate design does require three actuators, or four depending on design, due to actuation not being linked to the spine itself as it is in this model. The older iterations of the LIMS2 AMBIDEX wrist used a similar model to the alternative, having actuation be separated from the spine via wire actuators, though they used a mechanical spine of the same design as the segments shown here; the key point here is that the spine is not directly linked to the actuators. This design realistically works for only one to two segments do to its downfalls, whereas the more optimal design can go for many more segments. I have a feeling, due to the downfalls of this design, that this project would become a sunken cost fallacy even for the best designers, there's just no good way to optimize all the problems, whereas an alternative provides less issues from the start.
@@xaytana I agree with your point. Another design that can be an alternative for a quaternion joint used here is an Omni Wrist by Ross Hime Designs Inc.
Also looks like this could work as a sort of CV Joint with extended angular working range for delivering rotational force. Would need some serious reinforced alloy arms and tight tolerance bearings though. Absolutely love your videos, sir!
Ditto on that. CV joints and "virtual sphere rolling joints" solve the same problem, just for different applications. And given the variation in existing CV joints, modifying one of those may suit this purpose in a more robust fashion.
Add in a "Centering Spring". Just a piece of spring steel or coiled compression wire in the middle to give a restoring force to the assembly. Also look at joysticks and how they do it.
0:27 It's a bit like the cruciate ligaments in the knee. That's awesome. Or maybe the cruciate is more like the earlier example at 0:13 because the ligaments are like little cables or ribbons inside the knee.
Perhaps the use of worm/screw type gears on the motor side would work better? Also have the added benefit of reduced play in the driven structures to act as a better braking mechanism...?
I was thinking the same thing, and drive them using cables (like a Dremel flex shaft) so all the motors are in the base. That will also cut down the weight of stage 2 and 3 drastically.
wow, thats awesome! really nice work as usual! those rolling motions where both segments work together are really fascinating and elegant, like a dance. i know this was sponsored but it would be cool to see this design with stepper motors and belt reduction and maybe some carbon reinforcement of the floppy plastic links. maybe a soft robotics gripper at the end to turn it into an elephants trunk. :)
Really cool to look at, though it seems a bit inefficient to use three motors for two degrees of freedom at each stage. I wonder if there’s a way to recover that with a rotating DOF between stages that could come from a sum of all the motors or something with a differential like that.
Yeah he actually missed in his explanation that you can rotate+translate Z by rotating all servos the same direction. So the 3rd DOF is not something a traditional roboticist considers a DOF since its a mix of two DOFs. The 3 motors do all contribute their power to motion so having a 3rd one is not inefficient even if you constrain the calculations to not rotate the platform