Necessity for Multiple Categories of Modular Frame Types, As Well As Component Types

I'll start off pointing out that we have some really great ideas being shared here. I'll begin with the real life example brought up by DroidFreak: Phonebloks. I guess I missed the bus on that one, and hadn't heard of it until now (but then again, the concept was developed only two months ago). The idea behind Phonebloks will undoubtedly make its way into reality, because of its necessity. Phonebloks is an excellent example of an industry addressing the problem of not having modular frames. If you haven't watched the short video explaining Phonebloks, you must, because it's just that awesome:

How...cool...is that? What I find very interesting about all this is that the cell phone industry never globally standardized any frame type before, and the consequences of that have become so noticeable to the point where a group of people just like the members of Project FRAME had to organize a webpage to share ideas for a modular phone system. If I had heard of Phonebloks sooner, I would have mentioned it in my first post welcoming members to this project.

The concept I want to flesh out with this is the idea of an industry going from non-modular to modular. Let's be real; advanced robotics hasn't become much of an industry yet, but when it will, imagine how difficult it will be to make it modular if it does not start out as modular from birth. This is exactly why Project FRAME needs all the participation it can get, because advanced robotics is far more complicated than cell phones. Hell, it will use cell phone technology within it. If the future will bring us walking robotic vehicles that are each built for a unique frame, then the entire robot may be trashed if a single element stops working, just as the Phonebloks video above mentions. Or, if all you wanted was for the robot to move faster, why trash the entire robot when you can get hydraulics with higher performance?

Let's take the analogy of Phonebloks with Project FRAME a bit further. The video above clearly depicted the different component types that will be available for Phonebloks, such as the ones labelled below:

They also have a standard design for connection ports:

What makes it easy to accomplish, however, is that there is only one frame type. There is absolutely no need to have a different variation of the board that accepts a different type of blocks, because it's always going to be a phone. Advanced robots by no means will consist of only one frame type, so what needs to be done is generate a list of frame types that we can expect the industry to divide into.

An idea for an interesting frame type has been described in Bruce Jarel Chambers' first post, which involves having only three component types: The chassis mounted joint, the intermediate segment, and a terminus. The change I would make to this system is to have the chassis already outfitted with joints that connect to whatever element you want. I believe it would over-complicate element designs if the elements themselves had the mounting joint. From all my experience with robot designs, I have learned that the moving components should weight the least, to maximize motion flow. And, if these components are limbs, then all the major joints (shoulder and hip mountings) should be intrinsically based in the chassis. Let's take another look at Ambush from Real Steel:

Notice how the shoulder joints are heavily integrated into the chassis frame. The arms themselves seem to be attached without heavy amounts of joint connections, with nothing but a rotating cylinder connecting each arm to the shoulder joint. If the shoulder joint were actually to be a part of the arm, it would be much more difficult to attach it to the chassis.

Keeping it to Real Steel robots, the film presents us with another method of shoulder joint connection, used in the robot Noisy Boy:

After his arm gets broken off, we see that it was attached with a ball joint connection, but mechanical parts surrounding the ball joint are what actually powered the arm movements, just like how the human skeletal system is engulfed by the muscular system. It is interesting to note that the ball joint itself is mounted on a large lever, allowing the robot to perform motions such as shrugging (to defend it's head, as it is a boxing robot). Ambush seemingly can do all the same, but instead of the tension chords that Noisy Boy uses, Ambush just has a bunch of hydraulic cylinders. [Take a look at http://io9.com/5847701/designing-a-world-of-robot-fighters-the-concept-art-of-real-steel for a designer commentary on the Real Steel robots.]

So to conclude, one type of modular frame for advanced robots, as first suggested by Bruce Jarel Chambers and then modified here, could be a chassis featuring standard joint connections wherever necessary, and detachable limbs with detachable terminus components. I think it's pretty simple, and it could work nicely for smaller-scale up to human-sized robots, especially if they are humanoid. Larger machines would need a larger array of component types, a list that we still have to think up. I think a preliminary step for coming up with a very agreeable list is to first identify exactly what advanced robots will most likely appear in future industry. By this I mean what sort of basic structure, what sort of attached equipment, and what industry(s) they will serve. And of course, how all their modular parts connect together.

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.

Modularity in Media - A Modern Dream (and reality)

EDIT: I actually found a playable version (In Java 7 Update 25 or older) of the Stormrunner game if you're hankering to play it. It's a bit buggy, but then again it's 13 years old!

I remember long ago that one of my favorite flash games was the Lego Mindstorms RCX StormRunner game. The game, launched in the year 2000 (yeah, we're getting real nostalgic up in here) featured programmable RCX robots which you used to perform tasks in a hostile alien environment. More to the point, the RXC robots featured customizable parts. You could equip them with tracked or legged mobility, with a couple types of attachments, and with a variety of sensors. It was my first real encounter with robotics, modularity, and programming, and I loved it. Unfortunately, I was unable to find a playable version of the game when I searched today. I did find the intro of the game on the developers' website. Oh, 90's flash game animation. Gotta love it.

I digress - the point of this rambling nostalgic intro is that modularity is a compelling concept. Here's a much more recent example: PhoneBloks, the modular phone sensation that took phone-lovers' facebook feeds by storm this year.

It's an extremely similar concept to what we're proposing with Project FRAME, and it's gathered quite a bit of traction in public interest. So maybe we could do the same thing. We could call it RobotBloks... or RoBlox... whoops.

Which brings me to my point. Modularity is a huge deal in the public opinion. I'd venture that a large part of the reason Minecraft is so incredibly popular is that it's so modular.

 

Everything is a block, and all the blocks fit together in about every arrangement you can possibly imagine.

But that's not all. Oh no, you can't explore modern media without running into fantasies about or showcases of modularity. What shooter is complete without attachments for your guns? (Armor Mayhem, but that's another story)

And let's not forget that so many games include upgrades which modularly combine with your existing abilities.

And why stop at games? Movies have it too, like Hawkeye's arrows with modular heads in the Avengers

And let's not forget astromech droids like R2-D2 in Star Wars which seem to plug into everything.

The point is, people love modularity. It's a great tool for the user of any technology, enabling easy specialization and repair. It's already a reality in many places: desktop computers, cars, military weapon attachments, and - perhaps humblest of all - 3-ring binders.

With Project FRAME, we aim to help bring modularity to the next level with the promotion of advanced modular robots for a variety of applications. Join us, and help bring on the robotic revolution. The kind that helps humanity, not the kind where they take over the planet. Although that kind is always fun too.

Welcome to Project FRAME

"I was able to use the spin frame from Ambush and take the circuitry sleeve out of Noisy Boy, and it works."

                       - Max telling Charlie how he upgraded Atom, in the movie Real Steel.

        Ambush                                          Noisy Boy                                                                Atom                    

What is a "spin frame"? What is a "circuitry sleeve"? I Googled them both but got results only for variations of the words, and the results had nothing to do with robots. To conclude, these parts do not exist, at least with such terminology. We learn moments later that what Max actually achieved was the installation of voice recognition hardware into the deaf robot. The most interesting aspect of it, however, is that the new parts came from other robots.

I'm not here to deliver ideas from Hollywood, but we all know that machines are built from parts that are otherwise separate elements. What we seem to take for granted is that some machines in our world are made in a way where the composing elements can be traded out with completely different parts. I don't blame anyone; it's not very normal to think about how a Honda Accord with a V4 engine can so easily be refitted with a V6 engine. The thought only comes to mind when you actually want to modify the car intentionally.

Then there's the concept of adding parts without removing others, because the frame of the machine has vacant connection ports that have been standardized to allow any of a full set of custom parts. Take for example the pylons on fighter jets, such as the common F-16. Let's see what any of the pylons on the wings can hold...external fuel tanks? air-to-air missiles? laser guided bombs? Oh, I know - an advanced navigation system! How about at least one of each all at once? We have enough space for it, after all!

Maverick Missiles

But wait...it seems other fighters can use the exact same attachments that we just put on our F-16. Observe: the AGM-65 Maverick air-to-ground missile on both the F-16 and the F-15:

       F-16 with green Maverick missiles                                                 F-15 with grey Maverick missile

Keeping to the theme of military, let's look at another globally standardized frame: gun rails. Specifically the Picatinny rail system (which has been standardized, unlike the Weaver rail system). Any guns with this rail system can be fitted with the same scopes, holographic sights, flashlights, laser pointers, grips, grenade launchers, etc, no matter the gun.




      [A slice of a gun rail mount]

The concept being addressed is known as modularity. I'm sure we all have a strong understanding of what this means, but let's review the technical denotation of the word, pulling definitions from several dictionaries:

mod·u·lar·i·ty:

a) "The degree to which a system's components may be separated and recombined."
b) "Designed with standardized units or dimensions, as for easy assembly and repair or flexible arrangement."
c) "The use of individually distinct functional units, as in assembling an electronic or mechanical system."
d) "Having parts that can be connected or combined in different ways."

Excellent. Now we know four different ways to explain the meaning. Now that this has been established, allow me to explain the real juice of Project FRAME in terms of modularity. The fact that devices such as cars, jets, and computers have very modular construction is largely due to them having rather simple and distinct mechanical parts. A car has four wheels and an engine, with axle connections to transfer mechanical power. A jet has its engine as well, wings with tilting flaps powered by hydraulic cylinders for steering, and pylons that mechanically hook and unhook to attach devices or release missiles. Computers have no mechanical parts, save the cooling fan and the DVD drive. When the complexity of the mechanical parts is minimized, the system seems to enjoy wider modular capabilities, because it is much easier to globally standardize port attachments for hard-drives rather than axle connections for car engines.

Now we make way for the overarching purpose of Project FRAME. After the discussion about mechanical complexity limiting the global standardization of a modular frame, imagine the standardization of a humanoid robot frame. Or a quadrupedal vehicle frame. Or a robotic exoskeleton frame. All such machines are extremely complex mechanically, and it does not seem that they will be put into modular frames without taking specific efforts to do so. Cars for example became modular out of necessity and the ease of achieving it, whereas advanced robotic platforms are just far too complex to assume that they will naturally fall into standardized frames, a phenomenon exhibited by the auto industry. In fact, the advanced robots that do exist today are made only by single companies, organizations, or universities, and their design schematics are not being shared on the grounds of copyrights. You see, when you make something so complex and unique, yet it serves a difficult purpose, you don't really want to show how all the parts are connected together. You want people to come back to you to modify the machine after they buy it from you, like monopolization. What needs to be done is to form some sort of global agreement - among the companies of the world - to manufacture advanced robot elements that fit together no matter where the parts are from, or even how they are designed. The point is that their connection ports have to be exactly the same. A robotic claw with four fingers should connect to an arm where a drilling tool can be mounted instead, for example. Hydraulic cylinders that power heavy joints should be available at varying performance levels, differing by the amount of energy they consume. A vehicle that moves via caterpillar tracks should have the flexibility to have the tracks replaced with a multi-legged mobility platform. Now think - how on Earth will robotic machines become modular all by themselves? The answer is they won't. It happened to cars because it was easy. It happened to planes because it was easy. It happened to computers because it was like making pieces of a jigsaw puzzle. But advanced robots? Not really a homemade-jigsaw-puzzle type of thing. You need the manufacturers to agree that all variants of this element will connect this way, and all variants of that element will connect that way.

Now comes the creative part - what kinds of elements are we talking about exactly, and how should their connection ports be designed to prepare for global standardization? This is where Project FRAME really comes in. The discussion team of Project FRAME essentially has two goals: 1) to determine how to get companies to agree on advanced robot frame standardization, and 2) to present ideas for standard frame designs. I urge anyone even moderately interested to leave comments, or better yet to become an author of this blog by messaging me (moshibadalov@ymail.com). Project FRAME is designed to be a collaborative effort. The transition to the next generation of robotics technology will simply not happen if every organization is making separate robots that only they can fix and upgrade. Advanced robots will become far too expensive as the market becomes monopolized, which will inevitably happen unless a global effort is made to make a modular system, or systems.

To finish off, here's some food for thought. Compare the images generated by completely different artists. You may recognize the first one: