Sunday, March 18, 2012
I want to say a little more about other uses for PRT technology besides moving people. Freight has long been contemplated, and looks attractive as an after-hours revenue generator, but has the same critical mass problem as carrying people, only more so. Only after there is an extensive network is it of much use. In the last post I said that the concept of SMART (Standardized Multi-axis Automated Rail Transport) applied to more than Automated People Movers. Well, it also applies to more than freight. It applies to anything and everything that needs to be picked up at point A and sent non-stop to Point B within a matrix of possible origins and destinations that are most efficiently accessed using 3D routing. A prime example is manufacturing.
The images above represent a greatly simplified assembly line. On the left six parts (in red) are added to a product moving along an assembly line (blue) in various stages of completion. The right side shows what happens if the product comes in just three variations. No extra parts, just some hypothetical process variations at the points where the blue lines diverge. Now, when same parts are added, paths must cross. If this is done on an actual work floor, this crossing is quite literal, causing someone to wait, go over or go under. This situation is usually handled by stockpiling supplies, so such crossings are less frequent. But consider how this solution is analogous to city buses. They can move many passengers through an intersection at once, but are big and cumbersome. Of course we know what happens beyond a certain critical mass. At some point, anything other than full-speed, non-stop movement will create a cascading gridlock effect. If the parts are bulky, stockpiling them may be impractical in any case. So with that in mind, consider the illustration again, but this time with dozens of variations, instead of three, and many dozens, or even hundreds of parts, instead of six. This makes routing and staging parts and positioning work areas a real nightmare. It would take me a week just to draw it!
Some years ago, back when CNC was a brand new thing, I was given a tour of a cabinet factory. The main feature that you couldn’t ignore was that the place looked like it was designed by an insane ski-lift maker. There were cabinet pieces floating past all over the place, some hanging on hooks, and some in baskets. These were all going in different directions, all guided by long moving chains hanging at various levels from a superstructure attached to a very high ceiling. It was quite a sight! I think cabinet making is a simple enough process that hopefully it will help me illustrate my point.
The cabinets largely started as 4x8 foot sheets of material. You see, the secret to cabinets is that there are very few changes in saw settings. Be it a 48” or 24” cabinet, the sides, for example, are the same height and depth. And since most dimensions are repeated again and again, it only makes sense to set various saws one time and cut large numbers of pieces. And so it was in this factory. Material came in on forklifts and was rip-cut to various standard widths. Then those pieces were cut to lengths for the various sizes of cabinets. Most of the pre-finished pieces would go directly to an assembly area that was just for one or two specific cabinets. So there were many assembly areas. But cabinets come in various paint or stain finishes. All face-frames and doors and drawer fronts for a specific order, therefore, had to go through a finishing process. So they needed to be collectively detoured to a finishing and drying area and then re-sorted by sizes and sent to their respective assembly areas where they would join the prefinished parts and preassembled drawer boxes, which would be made still elsewhere in the factory. After assembly, the finished cabinets would be push on roller tables to a nearby boxing station, and then grouped with other sizes to make up an order or truckload. Cabinets come in four or more standard sizes, along with corner units, sink bases, and banks of drawers and all sizes have corresponding upper cabinets. And if all of this isn’t enough, let’s not forget the number a door styles that are typically offered!
Now consider all of this this against the illustration, and you will appreciate the problem. There were many dozens of work stations, some organized by component, some by cabinet size, some by finish, and some by purchase order. The only way to move things around efficiently was to have skyhooks that would whisk parts to the proper station, and do so without clogging up the shop floor. Cabinet parts would go around and around until someone took them down.
The same root causes of traffic on streets are the dilemmas that the cabinet factory suffered. These are generic problems that come from conflicting directional movement on a 2D (ground level) surface. As I think I may have mentioned in a previous post, the same problem extends all of the way down to circuit board and even computer chip architecture, and I have been told that there are aspects of graph theory that are used to help solve these difficult issues. Planning a factory, it seems, is a real science. I read somewhere that the Chevy volt assembly line is over twenty miles long. They are probably counting production lines for the sub-assemblies, but still this gives you an idea of the scope and complexity of this issue.
PRT grid would be used to prioritize pick-ups and deliveries around the factory floor. Every stockpile of pieces or sub-assemblies at every workstation would be automatically replenished, and then the resultant parts taken away to the next staging area. PRT traffic management, it turns out, is very close to industrial process management.
We are in the early stages of the robot revolution, and most are still essentially mechanical arms bolted to a floor. Any work piece has to come to them. Lately, however, some are becoming more mobile. Gaining popularity is what is called a gantry robot, because it can move itself along an overhead beam. But these are still tethered by power and data cables. While these robots are handy for assembly, because they can fetch parts, they can only go so far. The next logical step is the untethered “railbot.” This has not already happened, in part, because the key to maximizing the effectiveness of such a system is to make it route-switchable, electrified rail, but still with full data transfer capabilities. (Gee, now where have we seen that problem?) Make no mistake, this is coming to manufacturing. Non-switching, somewhat limited versions of such overhead “railbots” are already used to lift and tilt cars during assembly, as shown in the picture below.
Refrigerated warehouses already use similar machines to fetch frozen foods without requiring humans to work in the cold. The PRT maker 2getthere is a subsidiary of company that makes self-navigating forklift type vehicles. Making a grid over the factory floor and extending track all of the way to remote warehouses promises to make an extremely efficient and versatile manufacturing platform. Remember, the overhead bogies can either carry materials or have their own robot arms or both. As long as the bogie can be made to precisely locate itself and then lock on to the track, it’s suitable as a base for a robot arm. In the case of the cabinet shop, for instance, the assembled cabinet faces would be not just carried into the finishing room, but held for spraying and delivered to the drying area without human intervention. We are talking about an amazingly flexible system here.
The holdups to deployment are very similar to obstacles facing PRT. Like vehicle makers, robot makers do not really want the business of constructing track or controlling traffic, since that would demand a totally different, more locally oriented business structure. And track fabricators certainly don’t want to get into the robot or vehicle business either! Some great ideas just don’t fit with existing business structures.
That brings us to a point about the “S” (for “Standardized”) in SMART. One key is to have a standard track-to-vehicle interface, so these businesses can concentrate on what they do best. Currently, assembly line equipment is generally a huge, one-of-a-kind investment that may well go to the scrap heap when it is no longer needed. Meanwhile, product life cycles are, in case you haven’t noticed, getting increasingly short. A modular solution can be reused, reconfigured, and even eventually sold off at a reasonable price, like any other equipment. But modular pieces need to fit together, and that means a degree of standardization.
There was a time when getting a computer meant special rooms and teams of engineers to hook everything up. Over time, the system evolved into a something much more modular, and therefore flexible and simple. What geeks call “plug and play” is, no doubt, the destiny of automated manufacturing as well. It just happens that we, in the world of PRT, have long contemplated and largely solved the problems that they, in manufacturing, are just starting to understand that they have. And did I mention warehousing and package sorting?
When I talk about needing non-profit organizations, government grants, academia and corporate partners, it should be kept in mind that there is a lot of benefit to taking part in this transition. Self-navigating, non-stop 3D transportation is the ultimate in efficiency, both in terms of time and energy consumption, and this, in one way or another, will always mean profit. This application, unlike PRT, is for a closed, private sector application where there are a huge number of companies that might benefit. When I talk about an NPO umbrella for all of this, let me just say this: (I would refer the readers to the previous post for background)
The companies who would benefit the most from this technology need some kind organizational framework that enables them to participate and keep abreast of the work that others are doing. The tax exempt part of this really isn’t the point for these players, they can write everything off anyway. But I can see absolutely no reason to make this framework anything but an NPO, especially early on. This is not really anything revolutionary. The world is awash with such consortiums, and it would, in practice, be nearly impossible to equitably divide up any profits anyway. (An example of a for-profit consortium that had profit dividing problems is the original Airbus, and an example of a large NPO type consortium for advancing technical expertise is Sematech). This particular endeavor, I might add, also has a lot of “green” and “public good” aspects to it, and so would be well situated to raise money from individuals, charitable organizations and governments, not to mention possible participation from universities… I mean we ARE talking about saving the planet, after all…
And yes, I realize that this whole thing, at this moment in history, is an audacious pipedream. I also know that every revolution has to start somewhere. For lack of a consortium, I am, at least for now, willing to do its work singled-handedly, at least in respect to exploring the pros and cons of various design choices. A consortium of one!
I guess the bottom line here is that the ability to “airlift” many objects, non-stop, in a coordinated way, is a very big deal in many human endeavors, not just people. Robots are becoming more mobile and modular every day. So remember, you heard it here first: Railbots are coming to get you! (at your local station, of course!)