Cent per actuator robotics


I think the idea then is to embed platinum black into the threads and use methanol vapor to cause a heating reaction. I actually got that heating idea from TV here.
The energy density of doing that is much higher than rechargeable lithium batteries. Probably a little less energy dense than fats but the conversion efficiency into mechanical work could be better than the human body achieves. I guess you could fuel up a humanoid robot for a couple of weeks of continuous activity.

Wait, where did the “cent per actuator” part come from? It seems like these would still be pretty expensive due to how new the technology is, and there doesn’t seem to be enough information in that article to build the muscles without having a subscription to something.

However, the idea of a lot of cheap actuators got me thinking. There should be other ways to do it, and I think looking at old technology could help when trying to make things cheaper.

A couple methods I thought up are:

  • use a pulley system with a 1$ motor attached to each pulley
  • Standard pneumatics/hydraulics, given enough actuators (Unfortunately strength/response speed for larger ‘muscles’ is limited, since multiple muscles are driven by one pump)
  • getting a coil winding device and a magnetizing machine and using them to make a bunch of custom linear motors
  • 3d print screw roller linear actuators using cheap materials like PLA and connect them to hobby motors

Or, if you’re making something large that flies, there are some cool new thrusters that are actually an old technology:

and, on the sensor side, in addition to using voltage/current sensors/encoder wheels on those actuators, you could:

  • buy a multitouch track pad
  • buy some temperature sensors, or just regular transistors if you’re good with electricity and don’t need high accuracy from one sensor (you don’t, because you have many sensors).
  • get gyroscopes, accelerometers, and magnetometers
  • buy a lot of 1$ microphones and make a sonar sensor using triangulation (tetrahedulation?) and other machine learning technoques multiple times and combining to find the locations of the source noises
  • print out capacitive sensors using Hilbert’s space filling curves and a modified 2d or 3d printer and send high frequency AC through each capacitor to sense bending.
  • use spectroscopy to identify a few chemicals as long as they’re simple enough

It’s hard to find the cheaper, less accurate versions of some of those technologies, since most industrial applications tend to use one thing that matches specifications instead of many that don’t, but it’s not impossible. I listed the methods that should be easiest to get started with at the top of each list. Since each of these use many sensors/actuators, they should work well with HTM.

Nylon thread doesn’t cost much. You have some engineering choices about how to add electric resistance heating. Vacuum deposition metallization of the nylon thread would not be expensive.
There are other technologies to make the thread conductive as well. If it’s not a cent per actuator I’m sure you will be able to buy a pack of 10 small actuators for say $1.29 retail in a few years time. If you start using methanol vapor the physics gets complicated with gas diffusion rates to calculate etc. You would struggle to optimize the actuator for speed.