4 ms·

Could you elaborate a bit more on “formalize mechanical devices”? That has really piqued my interest especially since you mentioned dance choreography as an exa

Could you elaborate a bit more on “formalize mechanical devices”? That has really piqued my interest especially since you mentioned dance choreography as an example. Do you have any examples/prior art you found interesting? One of the topics I occasionally like to think about is how to formally describe manufacturing steps in some structured/programming language. Maybe your program could be it. Would love to hear other folks’ thoughts on this too!

felipereigosa 6y ago
Well, I was really into formalizing stuff when I was in college, you know proving that a cryptographic algorithm was correct by formalizing them as black boxes talking to each other, that sort of thing. Then I just started thinking about how to formalize mechanical things. One of the very things I remember is coming to the conclusion that all motions of a machine can be implemented as a combination of linear and rotational motions (and that's where the two things that move on mockmechanics, the track and the wagon, come from), but I don't know if that is true. But I don't really have any prior art, I had a 2d version that didn't work years ago but I think I lost it.
- orbifold 6y ago
  You should check out Shannon’s work on the differential analyser if you haven’t already. Before computers came along they used mechanical machines to solve differential equations etc. There are more papers with beautiful drawings of mechanical machines from that era.
- Joker_vD 6y ago
  I think the prior art you're looking for kinematics and robotics... and I distinctly remember that e.g. for multi-joints robotic arms you can actually use methods from the algebraic geometry instead of calculus (so you solve polynomial equations instead of differential equations). Also, no one in this thread mentioned Opus Magnum from Zachtronics which has some quite neat mechanical movement in it.
  - felipereigosa 6y ago
    I just saw the intro video for Opus Magnum and it looks really cool.
bigiain 6y ago
> Could you elaborate a bit more on “formalize mechanical devices”? It made me remember yesterday's post about SparkFun's new Al La Carte thing ( https://news.ycombinator.com/item?id=24920043 https://news.ycombinator.com/item?id=24920043 ) - and now I want a service that lets me design a machine in this using it's "standard parts", and then order a physical version of it :-) I wonder if some combination of lego/mechano style components, 3D printing, and magnetic electrical connections for batteries/switches/leds/motors could work? (Somewhat like this: https://www.pinterest.com.au/pin/178314466469771331/ https://www.pinterest.com.au/pin/178314466469771331/ which I had as a kid way back...)
- felipereigosa 6y ago
  You know I had actually considered this exact thing you want a while back as a way to move forward once I had the program working. You could design all the graphs, logical connections on the computer then load them and just snap all the parts together. The main drawback I encountered then was how to implement the wagon since it can move in an arbitrary track path circuit. I'm not sure how that could be done. Another idea that's a little out there is that if someone could implement the basic building blocks using molecular parts, then maybe that's how we build molecular machines!

felipereigosa 6y ago

Well, I was really into formalizing stuff when I was in college, you know proving that a cryptographic algorithm was correct by formalizing them as black boxes talking to each other, that sort of thing. Then I just started thinking about how to formalize mechanical things. One of the very things I remember is coming to the conclusion that all motions of a machine can be implemented as a combination of linear and rotational motions (and that's where the two things that move on mockmechanics, the track and the wagon, come from), but I don't know if that is true. But I don't really have any prior art, I had a 2d version that didn't work years ago but I think I lost it.

orbifold 6y ago
You should check out Shannon’s work on the differential analyser if you haven’t already. Before computers came along they used mechanical machines to solve differential equations etc. There are more papers with beautiful drawings of mechanical machines from that era.
Joker_vD 6y ago
I think the prior art you're looking for kinematics and robotics... and I distinctly remember that e.g. for multi-joints robotic arms you can actually use methods from the algebraic geometry instead of calculus (so you solve polynomial equations instead of differential equations). Also, no one in this thread mentioned Opus Magnum from Zachtronics which has some quite neat mechanical movement in it.
- felipereigosa 6y ago
  I just saw the intro video for Opus Magnum and it looks really cool.

bigiain 6y ago

> Could you elaborate a bit more on “formalize mechanical devices”? It made me remember yesterday's post about SparkFun's new Al La Carte thing ( https://news.ycombinator.com/item?id=24920043 https://news.ycombinator.com/item?id=24920043 ) - and now I want a service that lets me design a machine in this using it's "standard parts", and then order a physical version of it :-) I wonder if some combination of lego/mechano style components, 3D printing, and magnetic electrical connections for batteries/switches/leds/motors could work? (Somewhat like this: https://www.pinterest.com.au/pin/178314466469771331/ https://www.pinterest.com.au/pin/178314466469771331/ which I had as a kid way back...)

felipereigosa 6y ago
You know I had actually considered this exact thing you want a while back as a way to move forward once I had the program working. You could design all the graphs, logical connections on the computer then load them and just snap all the parts together. The main drawback I encountered then was how to implement the wagon since it can move in an arbitrary track path circuit. I'm not sure how that could be done. Another idea that's a little out there is that if someone could implement the basic building blocks using molecular parts, then maybe that's how we build molecular machines!