4 ms·

We need a breakthrough in low-speed, high torque, lightweight motors and geardowns. Basically if we can get economies of scale up on BLDC motors with higher num

by kortex 7y ago

We need a breakthrough in low-speed, high torque, lightweight motors and geardowns. Basically if we can get economies of scale up on BLDC motors with higher numbers of poles (think hoverboard motors but lighter), and figure out how to mass manufacture harmonic drives, that'll open up the door to all kinds of low cost robotics. Odriverobotics already has brought down the barrier to BLDC servomotor control. Or this thing: https://ieeexplore.ieee.org/document/8613852 https://ieeexplore.ieee.org/document/8613852

craftinator 7y ago
Two thoughts: 1) I haven't seen any clear literature on efficiency of affordable servos. It seems like existing technology must be very efficient, given the pretty uniform torque/power ratio I've seen across the servo market. 2) One of the biggest financial roadblocks I've run across in robotics isn't about the torque available in affordable servos, it's about the ability to get force feedback. Servos/motors are useless for the large set of tasks requiring force feedback, without the additional complexity of building a current sensor and calibrating it. We need a cheap servo with integrated force feedback and an easy means of autocalibration (from a quick search, seems like this feature usually doubles or tripples the cost of an identical servo).
- kortex 7y ago
  Torque/efficiency is still a big issue. A coworker and I put together all the math and specs for a robot which could "cook breakfast", just to see what a PoC would take. For anything approaching human arm levels of strength, in that form factor, you are looking at SWG servos. These are nominally 70-80% efficient, but that drops after a few thousand cycles, if you operate in higher loading regimes [1]. so cooling/maintenance is super important. Planetary gears are also not great efficiency-wise and are heavier. Force feedback is also super important and challenging. We just need like, synthetic actin :) Disney is actually working on that [2]. yeah it looks super inefficient but so were vacuum tubes before transistors. 1] https://eeepitnl.tksc.jaxa.jp/mews/EN/19th/text/206.pdf&ved=2ahUKEwiHkd2B57HnAhVZaM0KHU_xAKUQFjAPegQIAxAF&usg=AOvVaw2F4XHlavPrQQiyKJMZSuwt https://eeepitnl.tksc.jaxa.jp/mews/EN/19th/text/206.pdf&ved=... 2] https://studios.disneyresearch.com/wp-content/uploads/2019/03/High-Performance-Robotic-Muscles-from-Conductive-Nylon-Sewing-Thread-Paper.pdf&ved=2ahUKEwjG-oCP6bHnAhXBHM0KHeT7B4YQFjAZegQIBhAF&usg=AOvVaw1T86dwb46W3evrawujoK5n https://studios.disneyresearch.com/wp-content/uploads/2019/0...
  - craftinator 7y ago
    Interesting! Thank you for all the great information. I'm opening those links from a mobile app, and both ended up as not found (not 404, but as the sites internal unrecognized URL page). Are they live for you?

by kortex 7y ago

craftinator 7y ago

Two thoughts: 1) I haven't seen any clear literature on efficiency of affordable servos. It seems like existing technology must be very efficient, given the pretty uniform torque/power ratio I've seen across the servo market. 2) One of the biggest financial roadblocks I've run across in robotics isn't about the torque available in affordable servos, it's about the ability to get force feedback. Servos/motors are useless for the large set of tasks requiring force feedback, without the additional complexity of building a current sensor and calibrating it. We need a cheap servo with integrated force feedback and an easy means of autocalibration (from a quick search, seems like this feature usually doubles or tripples the cost of an identical servo).

kortex 7y ago
Torque/efficiency is still a big issue. A coworker and I put together all the math and specs for a robot which could "cook breakfast", just to see what a PoC would take. For anything approaching human arm levels of strength, in that form factor, you are looking at SWG servos. These are nominally 70-80% efficient, but that drops after a few thousand cycles, if you operate in higher loading regimes [1]. so cooling/maintenance is super important. Planetary gears are also not great efficiency-wise and are heavier. Force feedback is also super important and challenging. We just need like, synthetic actin :) Disney is actually working on that [2]. yeah it looks super inefficient but so were vacuum tubes before transistors. 1] https://eeepitnl.tksc.jaxa.jp/mews/EN/19th/text/206.pdf&ved=2ahUKEwiHkd2B57HnAhVZaM0KHU_xAKUQFjAPegQIAxAF&usg=AOvVaw2F4XHlavPrQQiyKJMZSuwt https://eeepitnl.tksc.jaxa.jp/mews/EN/19th/text/206.pdf&ved=... 2] https://studios.disneyresearch.com/wp-content/uploads/2019/03/High-Performance-Robotic-Muscles-from-Conductive-Nylon-Sewing-Thread-Paper.pdf&ved=2ahUKEwjG-oCP6bHnAhXBHM0KHeT7B4YQFjAZegQIBhAF&usg=AOvVaw1T86dwb46W3evrawujoK5n https://studios.disneyresearch.com/wp-content/uploads/2019/0...
- craftinator 7y ago
  Interesting! Thank you for all the great information. I'm opening those links from a mobile app, and both ended up as not found (not 404, but as the sites internal unrecognized URL page). Are they live for you?