Eight months ago he built a quadrupedal robot that could step sideways using three of them per leg. I’m not going to link that, you’ll have to find it from his YouTube page because you should look around.
Sorry for going off topic. "Electric Motor Scaling Laws and Inertia in Robot Actuators" by Ben Katz who designed the MIT Mini Cheetah in 2018 is very well known in the legged robotics community. His master’s thesis on actuator design is also widely referenced.
During the COVID period, some Chinese companies even sold variants of actuators inspired by the Mini Cheetah design.
Aaed Musa has also mentioned in some of his videos that his actuator designs were inspired by the Mini Cheetah actuator. Yes, His capstan drive video is especially impressive.
For example, in Aaed Musa’s video "I Built a Rubik's Cube Solving Robot" (https://www.youtube.com/watch?v=m0bMMALYMYk), he states in the description that the design was inspired by Ben Katz’s work.
I did a bit of a deep dive into motor control and actuators in building a two axis gimbal for tracking satellites with an RF antenna (with the eventual goal of building a mount for optical tracking).
Ben's vids were kind of mind-blowing for me at that time. I couldn't believe some of the control that was possible with relatively pedestrian electronics. Aaed's vids do a wonderful job of making it accessible in an applied way.
It's something I think a lot of the folks on HN would find interesting to tinker with. Nice mix of software and hardware that actually does work in physical reality. It also gives me a level of appreciation for the advances in humanoid robots that I don't think I would have had otherwise. (If you *do* get into it, I'd highly recommend getting into field oriented control with brushless motors and encoders. Small hobby servos are fun but they encapsulate a lot of the interesting parts and tend to have limited options available for things like the capstan vid linked above)
Thanks for this. I almost never have the patience for any videos, but this one hooked me and kept me engaged throughout. Worth it for that random "yo mama" joke alone.
This channel about retrofitting industrial robotic arm control systems is quite practical. Could always run the playback at 1.25 speed for slow talkers/edits =3
Same issue covered on HN a few weeks ago.[1] This one has more motor theory but less machine learning theory.
Too much gear reduction, and you can't back-drive or sense forces from the motor end. Too little gear reduction, and your motors are too bulky or too weak. Reflected inertia goes up as the square of the gear ratio, as the article points out, because the gear ratio gets you both coming and going. So high gear ratios really hurt.
Robots, like drones, need custom motors sized for the specific requirements of the joint. For a long time, the robotics industry was too tiny to get such custom motors engineered, and had to use motors designed for other purposes. This will become a non-problem as volume increases. Especially since 3-phase servomotor controllers, which drones need, are now small and cheap. They used to be the size of a paperback book or larger.
(I've been out of this for years. I've used hydraulic robots and R/C servo powered robots.
The newer machinery sucks a lot less.)
Reflected inertia does scale as the square of the gear ratio but it's a bit misleading unless you also consider the change in rotor inertia, which scales as a cube of the rotor radius (as the article points out).
The other side of the scaling laws say that motor torque scales as a square of air gap radius (roughly rotor radius), and output torque scales as linearly with gearing ratio.
When you balance these out, the reflected inertia depends on the inverse of power dissipated for a fixed output torque.
In an ideal world, your total reflected inertia is independent of the gearbox and largely depends on the motor fill factor and how hot you can run it.
Why can't we just dump massive currents into spring returning solenoids with ~5mm or ~1/4" range of motion, and amplify that motion through tendon systems for whole joint motion ranges?
Heat goes up with the square of current, so putting 10x the current to get 10x the force means 100x the heat.
Still, I think this idea is under-explored. There are probably applications for robots that move really fast, but only for a second before having to cool down.
TLDR; is that you need high current, meaning a lot of ohmic heating. With non-negligible back-EMF resulting in even more losses. Rotating motors essentially "lengthen" the travel of the "plunger" compared to linear motors.
I wonder if robots could be made to work better at cryogenic temperature, so superconductors could be used. The figure of merit would be much higher if resistance was zero. Or maybe this is another reason to want room temperature superconductors.
People gloss over peak torque versus thermal limits in robot actuators because stall torque even for a short burst will cook the motor long before copper embrittlement matters.
At cryo nobody sane is building robots unless the budget is a joke.
Even copper has vastly lower resistance when cryogenically cooled. It's not a bad idea for some applications, and water cooling is already a good way to increase power density.
I learned recently that the inductive heating coils used for metallurgy (smithing) are copper tubing with coolant flushing through them. The copper tries to heat up along with the bar you’re heating in the coil. Both from resistance and from radiative heating.
No, compliant mechanisms are important but the real gap in Robotics is perception (and no perception is not just Computer Vision).
There is such an insane amount of information richness in mammals and sensor specialization:
- Merkel discs
- Ruffini corpuscles
- Meissner corpuscles
- Pacinian corpuscles
- Muscle spindles
- Golgi tendon organs
- Nociceptors
And it's not the biology 101 types of sensors (in terms of variety of sources) but also the information density per square millimetre. It is orders of magnitude above anything that has ever been created with technology.
Once we leap above just Visual Servoing and start reaching sensory density and feature parity with human skin, joints, muscle, feet/hands, then we may start to see real breakthroughs.
https://youtube.com/watch?v=MwIBTbumd1Q
Eight months ago he built a quadrupedal robot that could step sideways using three of them per leg. I’m not going to link that, you’ll have to find it from his YouTube page because you should look around.
During the COVID period, some Chinese companies even sold variants of actuators inspired by the Mini Cheetah design.
Aaed Musa has also mentioned in some of his videos that his actuator designs were inspired by the Mini Cheetah actuator. Yes, His capstan drive video is especially impressive.
For example, in Aaed Musa’s video "I Built a Rubik's Cube Solving Robot" (https://www.youtube.com/watch?v=m0bMMALYMYk), he states in the description that the design was inspired by Ben Katz’s work.
Ben Katz master thesis, is worth reading: "A low cost modular actuator for dynamic robots" 2018 https://dspace.mit.edu/handle/1721.1/118671 and also has a good post https://robot-daycare.com/posts/2019-12-16-the-mini-cheetah-...
And also, The Rubik's Contraption (2018), 297 points, the work done by same author https://news.ycombinator.com/item?id=16561049
Ben's vids were kind of mind-blowing for me at that time. I couldn't believe some of the control that was possible with relatively pedestrian electronics. Aaed's vids do a wonderful job of making it accessible in an applied way.
It's something I think a lot of the folks on HN would find interesting to tinker with. Nice mix of software and hardware that actually does work in physical reality. It also gives me a level of appreciation for the advances in humanoid robots that I don't think I would have had otherwise. (If you *do* get into it, I'd highly recommend getting into field oriented control with brushless motors and encoders. Small hobby servos are fun but they encapsulate a lot of the interesting parts and tend to have limited options available for things like the capstan vid linked above)
https://www.youtube.com/@ExcessiveOverkill/videos
Too much gear reduction, and you can't back-drive or sense forces from the motor end. Too little gear reduction, and your motors are too bulky or too weak. Reflected inertia goes up as the square of the gear ratio, as the article points out, because the gear ratio gets you both coming and going. So high gear ratios really hurt.
Robots, like drones, need custom motors sized for the specific requirements of the joint. For a long time, the robotics industry was too tiny to get such custom motors engineered, and had to use motors designed for other purposes. This will become a non-problem as volume increases. Especially since 3-phase servomotor controllers, which drones need, are now small and cheap. They used to be the size of a paperback book or larger.
(I've been out of this for years. I've used hydraulic robots and R/C servo powered robots. The newer machinery sucks a lot less.)
[1] https://news.ycombinator.com/item?id=47184744
The other side of the scaling laws say that motor torque scales as a square of air gap radius (roughly rotor radius), and output torque scales as linearly with gearing ratio.
When you balance these out, the reflected inertia depends on the inverse of power dissipated for a fixed output torque.
In an ideal world, your total reflected inertia is independent of the gearbox and largely depends on the motor fill factor and how hot you can run it.
Still, I think this idea is under-explored. There are probably applications for robots that move really fast, but only for a second before having to cool down.
TLDR; is that you need high current, meaning a lot of ohmic heating. With non-negligible back-EMF resulting in even more losses. Rotating motors essentially "lengthen" the travel of the "plunger" compared to linear motors.
Cooling in general is not a bad idea to allow you dissipate heat as you push motors to their saturation limits.
Plus, cryo temps require a lot of power to keep thing cool and coper and iron embrittle so the forces acting on the winding could shatter them.
At cryo nobody sane is building robots unless the budget is a joke.
There is such an insane amount of information richness in mammals and sensor specialization:
- Merkel discs
- Ruffini corpuscles
- Meissner corpuscles
- Pacinian corpuscles
- Muscle spindles
- Golgi tendon organs
- Nociceptors
And it's not the biology 101 types of sensors (in terms of variety of sources) but also the information density per square millimetre. It is orders of magnitude above anything that has ever been created with technology.
Once we leap above just Visual Servoing and start reaching sensory density and feature parity with human skin, joints, muscle, feet/hands, then we may start to see real breakthroughs.