2D rail-frame based distributed modular pick-place assembly system
Basically macroscale prototypable instances of RepRec pick-and-place robots but with
★ a specific focus on 2D rail-frame systems as a base as the main scope narrowing characteristic
★ perhaps a slightly lesser focus on minimizing energy dissipative clip actions
★ perhaps a stronger focus on distributed system architecture
… for emergent self-replicative capabilities that a pure macroscale focus would lead to
★ a scale agnostic design so far that is possible/sensible.
… Physics change aware scale transposed prototyping inspired designs
… that hopefully still yield something useful for the macroscale context of:
… RepRec pick-and-place robots (RepRap) too.
Extensive intro side-nodes
Focus on 2D frame as a base
ReChain frame systems of RepRec pick-and-place robots
might include 3D spacetruss or spaceframe as structures
for the robootic system to move along too. Or might not. It is not specified for them.
See: Base Truss Crawler
This page here covers systems that have a stronger focus in 2D rail grids as a base.
This matches better to the scaling law of higher throughput of smaller machinery.
And is also likely better for ideas like the early nanosystem pixel (direct path).
As the flat thin 2D layout makes for good observability and debuggability.
Albeit more advanced nanosystems will eventually feature mechanosynthesis cores in assembly line assembly
making for a more 3D homogeneous base layer.
See: Deliberate slowdown at the lower assembly levels
Lesser focus on minimizing energy dissipative clip and high energy turnover actions
The Rebar chain tensioning applied in ReChain frame systems as elrments of RepRec pick-and-place robots
allow for minimizing the amount of high energy turnover actions.
This might well be massive over-engineering even for the nanoscale. Details below.
Thus putting a lesser focus on it here for the sake of simpler assemblies.
So long one can find good alternative means for connection.
See: Strength and stiffness preserving clip connections
Tolerable bending and snaps
Macroscale systems have low relative power losses in clicks compared to the drive systems baseline.
So no issue there aside wear from bending and unpleasant loud clanking noise from fully dissipative snaps.
Early nanosystems will operate with very slow motion. So no issue there either.
Aside crystolecule newton's cradle effect risks perhaps.
And even in advanced nanosystems in an assembly process every connection is usually just operated once.
So even if it gets quite hot it won't be able melt itself.
Much like rocket exhausts taking almost all their energy with them and thus not melting the nozzle.
The assembled product takes almost all the energy and is thus not melting the gem-gum nanofactory chip.
Still not tolerable bending and snaps
Notable exceptions to tolerability of energy dissipation in the assembly process
in future advanced high performance nanosystems would include:
★ Often actuated temporary holding clips (including the use of vdW forces as such) that stay in the factory.
– These may likely need a focus on energetic reversibility in gemstone based advanced productive nanosystems.
– And even the bending alone is severe might pose a problem as in the next point.
★ MHz frequency beding in drive systems like belts around wheels and large amplitude actuated flexures.
While the relative losses per cycle (Q factor) may be small,
– there are many cycles per unit of time and unit of volume, and more importantly even
– there is a huge energy turnover baseline due to the base materials being so extremely stiff and simultaneously superelastic.
… I feel the need to give a heads up to not get fooled by the scaling law of Lower stiffness of smaller machinery here.
… See: A better intuition for diamondoid nanomachinery than jelly
Bigger focus on distributed system non-monolithicity of self replicativity
All scales
There is absolutely no need to have full stack autogenous cycle of selfreplicativity,
The point is just to remove critical accessibility barriers. Not to demonstrate life like autonomy/autarky.
Nanoscale
At nanoscale pushing for compact self-replicativity seems highly detrimental to progress.
See: Why ultra-compact molecular assemblers are a bad and long outdated idea
Therefore highly distributed systems are a good focus for this context.
It would also be a natural choice to largely factor out electronic control in early systems.
Hugely reducing the remainder (of subsystem complexity) of what is necessary.
Macroscale
At the macroscale with a bit of a focus on co-usability for future nanoscale systems (scale transposed prototyping)
… it makes sense to factor functional units apart as much as possible
… generally factoring apart concerns (separation of concerns) should make design easy
… (helpful as a self recursive design of a pick-n-place robotic system assembling copies of itself is hard)
Smaller size (even still macrosacle) => distributed systems more viable:
Generally the smaller the (still macroscale) parts the better this distributed system approach should work.
As (beside assembly, material, and space costs)
part sizes, masses, and the according natural assembly frequencies seems to be a main reason for
why there are still no standard mechanical elements in analogy to the standard electrical elements we have.
See: Digital control over reversibly composable units of matter & MecCirc project.
For the mainstream RepRap movement the main (by now achieved) goal …
★ was to take out the cost (and thus accessibility barrier) to the complex shaped parts which made all the difference
★ was not to get to full selfreplicativity of all parts (except fro a small sub-movement)
– vitamins like electronics and metal parts where totally allowed
For an eventual RepRec movement the main (not yet achieved) goal …
★ would be to take out the cost (and thus accessibility barrier) of complex assemblies (currently done by hand or big factory)
★ would not be to get to a full stack self replicating system
– vitamins are still totally allowed, and heck, here include pre-made parts
… Pre-made by whatever means suitable. See page: Potential manufacturing methods for macroscale prototype parts
At the macroscale one is forced to relatively compact self replication (so far that is desired) due to …
★ operation frequencies being low
★ cost for the parts in material and manufacturing being high
★ reasonably available space being limited (especially in DIY RepRap like settings)
★ self mass load of the parts accumulating (unless one works in microgravity, hardly DIY accessible in the next few years put mildly)
★ materials lacking stiffness (especially affordably manufacturable ones like plastics)
Why this is not hypocrisy on avoiding ultra compact self replication
But isn't this ultra compact self replication after all and thus hypocrisy?
★ With the assemble of parts we are already at the second assembly level
… already assuming a supplied stream of a set of complexly shaped standard parts.
… replication is easier to get compact there especially when allowing to factor out electronics to "not part of the system"
★ I.e. This ultra compact partial self-replication at the second assembly level
… has hardly any resemblance to the idea of
… ultra compact nigh fully autogeneous self-replication of molecular assemblers at the first assembly level exclusively
Some basic system aspects
All unit positions and signal-line coupling states are always kept in machine phase.
I.e known positions. Nothing should ever free-wheel/free-slide around unconstrainedly.
Perhaps zeroing for damage repair as sole exception. The damage already sot one out of machine phase.
Unit-mover-units
No need for every unit to be a mobile "critter".
Actually only a single unit needs to be able to move about freely.
Here freely meaning 2D across the so far built rail-frame-grid.
All other units are carried around by one or more of this specialized "unit mover units".
The movements on this scale does not need to be particularly fast.
This is not part of the movement used for pick and place assembly.
With very few exceptions perhaps. Big pre-assemblies?
The unit mover units …
★ can move in 2D on the 2D rail-frame
★ are driven by signal lines in the rail frame.
Several types of clutches
There are at least two types of clutches.
★ One clutching to the 2D rail-frame structurally
★ One other type to clutch to mechanical signal (and/or drive) lines running through the rail-frame.
It is important to not conflate these two types as they serve orthogonal functions.
Signal line clutching patterns can establish different broadcasting groups.
(Groups as small as one for eventual debugging needs.)
Unit-mover-units operate the structural clutches of the units they transport.
For "designing against Murphies law" aim for designs that make physically impossible any clutching states
that would allow any part becoming free-wheeling/free-sliding.
Assembly stage units
These are not mobile on their own.
They need unit-mover-units to be moved about.
These need to be capable of quick motions.
So inchworming strategies are not an option for the motions here.
Multi phase drive lines with short stroke can be converted
to fast long range motions within the unit
One thing that is fundamentally wholistic and can't be broken up into smaller sub-units is the 6DOF motion stage.
There is the challenge of mechanical through joint threading (or RR/Mechanical through joint motion threading)
Threading through the RepRec poser which entails
threading through the RepRec positioner and esoecually challenging though the RepRec orienter.
some DOFs even threading through the RepRec end-effector.
There is the limited range fast moving vertical z-axis of the stage (part of RepRec positioner)
Building things up even higher would involve lifting up pre-assemblies from below then there
never to be touched again by the RepRec end-effector of the assembly stage units.
Eventually these bigger structures above cold become 3rd assembly level robotics.
Basically the a similar system again just bigger and above.
This should be a self-emergent process for nanosystems.
Questionably useful for macrosystems.
rail-frame extender units (possibly it's own unit)
Given a sturdy base (macroscale ceiling, floor, wall; nanoscale chip surface)
there's is no need for stiffening structures. This is usually a reasonable assumption to have available.
Without that complicatingly there is a need to go 3D for stiffening ind oder to be able to extend further laterally.
Mechanical signal & drive lines
Rotating axles severely lack stiffness fro longer transmission lengths.
Early nanosystems might have difficulties with rotationally symmetric parts.
Conventional roller chains have several issues.
They need turnaround sprocket wheels and are difficult to change in length.
Requiring mid chain element insertion/removal in a flexible chain and shifting at least one of the turnaround sprockets
Then one gets two chains spaced apart widely by the sprocket diameter.
Much simpler: Multi phase translatory reciprocating rods might be a good choice.
A sort stroke translatory reziprocating motion avoids large length overshoot into eventually adjacent domains.
Just like with roller chains one will want to minimize the amount of bearing contacts points to minimize on friction.
Macroscale: mass is a concern for these rods.
Nanoscale: mass is largely irrelevant, especially for early even slower moving systems
Control units & motor amplifier units
Maybe these two should be combined units
as most of the compute is supposed to be entirely outside the system.
Control units
Interface to the control electronics factored out for the system.
Connecting to the signal lines controlling their motions.
Motor amplifier units
- Macroscale pro: mass or the motors is not in the assembly moved about wiggling it
- Nanoscale pro: volume of the motor is not in the assembly stage massively increasing it's size
Parts magazine units
A two stage hierarchical architecture.
Every part type has a part type specific "magazine carrier".
The magazine units are have a (or are a) magazine for magazine carriers.
Alternativley one uses a generic adapter for each individual part.
Downside: This can't exploit natural compact stackings of some parts to get denser packings.
Parts magazine streamer units
Quickly delivering parts
★ from the parts magazine units
★ to the assembly stage units
★ or vice versa in disassembly
★ or between magazine units
Usage of attachment chains seems self-suggesting.
Parts magazine streamer units can chain and do streaming hand-over between them.
Note that there is an access speed vs capacity thing going on
in close analogy to the L1, L2, L3, … memory caches in CPUs.
One idea would be (un)fillng the magazine LIFO from "ground" level.
Same parts means LIFO does not matter in access time.
The problem that these units solve:
"assembly stage units" can't reach a lot of parts when they are placed inside their build volume
and carrying the whole units around by "unit mover units" to fetch new parts would be way too slow.
Well granted "unit mover units" could bring new "magazine units" as a middle ground solution.
Router units?'
Parts magazine streamer routing units as a sub-class here would be thinkable,
bus they'd need an active data stream then beyond the clock.
Ideally only the assembly stage and the unit mover inits need that.
The rail-frame-grid may have very limited capacity in the amount of mechanical signal & power lines.
Bonus: Specialized pre-assembly units (factorylets)
Assembly of modular struts, chains, rigid chains, atachment chains.
Parts in - "crank" - preassembled product out - or vice-versa.
Magazine carriers might be used directy from the stream of "parts magazine streamer units".
Beyond scope here
Part manufacturing units
- Macroscale see: Potential manufacturing methods for macroscale prototype parts
- Nanoscale see: Mechanosynthesis core
Future advanced desktop scale systems, not what this all here is about, would be petacore. See: Atom placement frequency
The architecture depicted in the figure here:
Early diamondoid nanosystem pixel (direct path)
Is actually very similar to the one described here.
With the main difference being that the first assembly level is included too.
The fist assembly level there being force applying mechanosynthesis in PPV.
There are according part magazines for the tool-tips and
vacuum lock-out units as a nano-physics specific ones.
The first assembly level is included even dominantly so.
It gets more hardware in ratio relative to the actual stick-n-place assembly robotics that is the focus here on this page.
This is …
★ because the crystolecule parts making process is much slower than the subsequent assembly of them.
★ a consequence of level throughput balancing.
★ a sensible optimization to do even in early nanosystems.
Note: There are no yet any specialized part pre-production streaming systems
for many hundreds of different part types each with dozens of stations.
These kind of optimizations are the ones only reserved for really advanced systems.
High level compute units
As stated earlier this is supposed to be
fully factored out for macroscale systems and early nanosystems.
All the described units here are supposed to carry just a few status bits at best.
Like the active broadcast group membership and such.
Note that there are no yet any specialized part pre-production streaming systems
for many hundreds of different part types each with dozens of stations.
These kind of optimizations are the ones only reserved for really advanced systems.