Difference between revisions of "Open loop control"
(→Related: added three relevant links) |
m (→Related) |
||
| (3 intermediate revisions by the same user not shown) | |||
| Line 24: | Line 24: | ||
Electron microscopy is sharing the destructiveness problem of X-rays. | Electron microscopy is sharing the destructiveness problem of X-rays. | ||
| − | In all cases, both light and matter-wave | + | In all cases, both light and matter-wave visual imaging, <br> |
both the generation and the detection seems not very miniaturizable. <br> | both the generation and the detection seems not very miniaturizable. <br> | ||
At least not down to scales similar to the nanoscale assembly chambers to surveil. | At least not down to scales similar to the nanoscale assembly chambers to surveil. | ||
| Line 43: | Line 43: | ||
But still sealed to [[PPV]] vacuum quality. <br> | But still sealed to [[PPV]] vacuum quality. <br> | ||
Transparent windows are not possible for advanced [[neutral matter wave microscopy]]. | Transparent windows are not possible for advanced [[neutral matter wave microscopy]]. | ||
| + | |||
| + | == Why to surveil the assembly process tactilely (and go beyond open loop control) == | ||
=== Fully integrated error detection beyond open loop control === | === Fully integrated error detection beyond open loop control === | ||
When going beyond open loop control at the nanoscale <br> | When going beyond open loop control at the nanoscale <br> | ||
| − | the most feasible error detection method is simple ''' | + | the most feasible error detection method is simple '''probing by touching.''' (testing for steric obstruction). <br> |
Is there part there where and when as it should be or not? What to do if exceptions are not met. | Is there part there where and when as it should be or not? What to do if exceptions are not met. | ||
| Line 57: | Line 59: | ||
it becomes increasingly possible to raster probing like imaging permanently into production systems. | it becomes increasingly possible to raster probing like imaging permanently into production systems. | ||
Not just for testing and debugging campaigns. | Not just for testing and debugging campaigns. | ||
| + | |||
| + | == Development & Final systems == | ||
| + | |||
| + | Development necessarily needs to go beyond open loop control. <br> | ||
| + | One needs to somehow "see" if the advanced [[productive nanosystems]] one builds actually do what they are intended to do. | ||
| + | |||
| + | Later unnecessary systems for observation can be stripped off again. | ||
| + | |||
| + | Related: [[Naked core]] and [[Legacy littered arrival result]] | ||
== Related == | == Related == | ||
| − | * [[Machine phase]] | + | * '''[[Machine phase]]''' |
* [[Tracing trajectories of component in machine phase]] | * [[Tracing trajectories of component in machine phase]] | ||
* [[Physical debugging]] / Analytics | * [[Physical debugging]] / Analytics | ||
Latest revision as of 17:51, 28 September 2022
Open loop control is control without feedback
The robot is not capable or equipped to "see" or otherwise detect
when its actions do not lead to desired/expected/nominal results.
Instead it just keeps on going doing what is does.
Totally oblivious and in total denial of what is going on.
Failures may lead up to self destruction of the robotic system
or sometimes to more or less spectacular and miraculous recoveries.
(wiki-TODO: add illustrations, two 3D printing failure modes, this is fine meme, ...)
Contents
Why not to surveil the assembly process ("visually")
Unlike in the macroscale one cannot just add cameras to detect and react on errors.
- Visible light has a way too long wavelength.
- Short enough wavelength light (X-rays) have way too much energy mostly penetrating not reflected and destructive.
Additionally the necessary computing power to "visually" surveil all assembly stations is not really possible.
Other means of imaging without direct physical include electron microscopy and advanced neutral matter wave microscopy.
Electron microscopy is sharing the destructiveness problem of X-rays.
In all cases, both light and matter-wave visual imaging,
both the generation and the detection seems not very miniaturizable.
At least not down to scales similar to the nanoscale assembly chambers to surveil.
One-off debugging campaigns
Gentle observation with advanced neutral matter wave microscopy should indeed be possible for
a singled out area of nano- to micro-machinery.
Even if strongly miniaturized (way smaller than room sized) such microscopes
will most likely always remain macroscopic in their size. br>
So there's no way to observe ALL of the internal nanomachinery "visually" in this level of detail during nominal operation.
One can only pick out an some areas of interest as statistical sample in the course of a debugging campaign.
The system musts be specially prepared to be open-space-accessible.
Like an open gem-gum factory cross section.
But still sealed to PPV vacuum quality.
Transparent windows are not possible for advanced neutral matter wave microscopy.
Why to surveil the assembly process tactilely (and go beyond open loop control)
Fully integrated error detection beyond open loop control
When going beyond open loop control at the nanoscale
the most feasible error detection method is simple probing by touching. (testing for steric obstruction).
Is there part there where and when as it should be or not? What to do if exceptions are not met.
Repeated equivalent testing for steric obstruction can increase reliability exponentially.
(wiki-TODO: Refer to Nanosystems section on that topic.)
This is obviously continuously extendable to a raster touch probing
eventually giving height-map images. Especially at larger scales (higher assembly levels)
it becomes increasingly possible to raster probing like imaging permanently into production systems.
Not just for testing and debugging campaigns.
Development & Final systems
Development necessarily needs to go beyond open loop control.
One needs to somehow "see" if the advanced productive nanosystems one builds actually do what they are intended to do.
Later unnecessary systems for observation can be stripped off again.
Related: Naked core and Legacy littered arrival result
Related
3D printers:
- Spaghetti failure mode
- Lost steps terrace failure mode
External links
- Wikipedia: Open-loop controller
