Modularity: a starting point

There are several ways to go about making something modular.  We can take cues from Lego, Picatinny, the US Air Force, Toyota, and the Electronics Industry.

Lego makes an excellent product wherein any component can attach to a mighty large percentage of all other components, and although anything can be done to a certain degree using Lego components, this sort of scale is completely impractical when you start talking standardized humanoid robots.

Picatinny takes the Lego approach to a very limited scope.  They do one thing, and do it very well.  Any Picatinny rail will accept any attachment designed for use with Picatinny rails.

The Air Force has managed to make their armaments completely interchangeable, as was in the previous post.

Toyota takes the biscuit for modularity.  Any Toyota-made or modified vehicle can be repaired or modified at any Toyota dealership.  This is made possible by the fairly narrow selection of parts that Toyota builds their vehicles with.  A Toyota-purposed leaf-spring will be one of 3 or 4 lengths.  Any Toyota-purposed shock will use the same mounting screws.  Any Toyota-purposed turbocharger kit will work with any Toyota-built engine.

USB.  Need I say more?  I'll just drive the point in by mentioning Arduino.

Taking the lessons learned from the above parties, one could build a limb system on a scale somewhere between Lego and Toyota, replete with perfectly standardized attachment points, USB based intercommunication mechanisms, internal Arduino-based brains, and resplendent in its inordinate number of Picatinny Rails.

This by no means is a final or optimal idea, but I was thinking a system in which each limb can be broken into varying numbers of sections comprising 3 types: chassis-mounted, intermediate, terminus.  The chassis-mounted section would be a joint, but each class (I'm thinking using 1/2 differences in connection parts) would contain a set of perfectly interchangeable parts in the intermediate and terminus parts.  Thus, a humanoid arm could be built using one chassis-mounted joint, one intermediate section, and one terminus.  A humanoid leg would be very similar.  A 3-segmented limb would have the chassis mounted joint, 2 intermediate components, and a terminus.  Proper engineering would also permit this to use Caterpillar treads.  The chassis-mounted joint would align into the position designated for treads, and the treads would attach as termini components.  Connection ports are as yet completely undesigned (have fun designing these, I beseech you), but I was thinking something where each of the 1/2 inch separated classes use something that looks completely different.  (more fun to be had)

Chassis could be dealt with in a separate, but also modular fashion.  I support the idea of putting forth ideas on how exactly how to do this.  I was thinking something along the lines of modular triangle frames.  This idea still needs refining, forging, and tempering, so could be used as inspiration or a starting point. (More fun still?)

Also, open source should be considered.  It has the benefits of preventing proprietary parts, and produces amazingly standardized and modular hardware and software.

Another question to be asked regards whether to use existing standards wherever we can or to invent a comprehensive standard so as to optimize the system.  I think that using arduino boards and USB cords and standardized bolts would be optimal, as the infrastructure to support these systems is already in place.  However, for such applications as giant industrial machines, individual companies should be left to invent their own standards.

The software should all be written in Python, and the guys who maintain and update Python should put all important functions into their import library.  I will embrace other languages upon a good defense.

Honestly, I invite you to design stuff, but only if you would enjoy doing so.  If you are feeling particularly particular, you could even draft up several comprehensive standards using any combination of existing and invented standardized components.