Design Notes
Ross Gayler
2021-06-17
Last updated: 2021-06-18
Checks: 6 1
Knit directory:
VSA_altitude_hold/
This reproducible R Markdown analysis was created with workflowr (version 1.6.2). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.
The R Markdown file has unstaged changes.
To know which version of the R Markdown file created these
results, you’ll want to first commit it to the Git repo. If
you’re still working on the analysis, you can ignore this
warning. When you’re finished, you can run
wflow_publish to commit the R Markdown file and
build the HTML.
Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
The command set.seed(20210617) was run prior to running the code in the R Markdown file.
Setting a seed ensures that any results that rely on randomness, e.g.
subsampling or permutations, are reproducible.
Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.
Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.
The results in this page were generated with repository version d8b9d53. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.
Note that you need to be careful to ensure that all relevant files for the
analysis have been committed to Git prior to generating the results (you can
use wflow_publish or wflow_git_commit). workflowr only
checks the R Markdown file, but you know if there are other scripts or data
files that it depends on. Below is the status of the Git repository when the
results were generated:
Ignored files:
Ignored: .Rhistory
Ignored: .Rproj.user/
Ignored: renv/library/
Ignored: renv/staging/
Unstaged changes:
Modified: analysis/design_notes.Rmd
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
These are the previous versions of the repository in which changes were made
to the R Markdown (analysis/design_notes.Rmd) and HTML (docs/design_notes.html)
files. If you’ve configured a remote Git repository (see
?wflow_git_remote), click on the hyperlinks in the table below to
view the files as they were in that past version.
| File | Version | Author | Date | Message |
|---|---|---|---|---|
| Rmd | d8b9d53 | Ross Gayler | 2021-06-18 | Add DFD for Design 01 |
| html | d8b9d53 | Ross Gayler | 2021-06-18 | Add DFD for Design 01 |
| Rmd | 0b7e1d4 | Ross Gayler | 2021-06-17 | Publish initial docs |
| html | 0b7e1d4 | Ross Gayler | 2021-06-17 | Publish initial docs |
This notebook records the design considerations behind implementing altitude hold in VSA.
1 Strategy
Limit the initial problem to altitude hold because this is one-dimensional (as opposed to fully general flight control). This is done to simplify the problem as much as possible.
Treat the VSA components as drop-in replacements for standard control circuitry. This reduces the probability that we will hit an obstacle requiring us to completely reconceptualise how altitude hold is done. On the other hand, it means that we will parallel the classical altitude hold computation, so we may miss seeing some (hypothetical) VSA-centric solution that has no classical analogue.
2 Design issues
2.1 Interfacing
How do we interface between the VSA components and the classical components?
VSA values are very high dimensional vectors.
The values in the classical implementation of the data flow diagram are very low dimensional vectors (typically scalars).
The surrounding machinery of the multicopter simulation assumes that the inputs to and outputs from the altitude hold are classical values. Therefore interfaces are need to convert between the classical and VSA representations.
The interfaces between the classical and VSA are essentially projections between low and high dimensional vector spaces.
2.2 Representation of scalars
Magnitude vs. spatial
Range limits
Noise/resolution
2.3 Temporal issues
Standard VSA is point-in-time/static (some implementations may be inherently temporal)
Embedded in discrete-time simulation. Compatibility of time scales.
Temporal resolution of signals?
Low-pass filters on everything?
2.4 Initialisation
- Classical implementation has explicit initialisation. Do we need special VSA circuits to guarantee orderly turn-on? Could we have an “always-on” VSA design?
3 Design 01 - No VSA
This section deals with the altitude hold controller before the introduction of any VSA components.
3.1 Data Flow Diagram
The data flow diagram represents the desired calculation for altitude hold.
The data flow diagram is derived from
AltitudeHoldPidController.Each vertex in the diagram represents a value and the computation that calculates the value.
Each directed edge in the diagram represents a value flow from one vertex to another
Think of the data flow diagram as specifying an electronic analog computer where the vertices correspond to function blocks and the edges correspond to wires.
Data flow diagram supplied by Simon D. Levy
What is the difference between the rectangular nodes and the unboxed nodes? Unboxed appear to be arguments to
__init__, so does that mean they are constant parameters over the life of the system?For each value (node) , what is it’s type? I presume they are all real scalars.
For each value (node) , what is it’s range? A physical implementation necessarily has limits. This may render \(k_{windup}\) and
constrainredundant.For each value (node) , what is it’s value resolution? This is equivalent asking the noise level that can be tolerated.
For each value (node) , what is it’s temporal resolution? This is equivalent how rapidly we expect the value to change. Some nodes may be constant parameters. For time-varying signals do we expect them to alternate between the extreme values on successive samples.
A related point: this is a discrete-time simulation, is the time step fixed? What is the time step? The simplest VSA approach is point-in-time. However, it may be possible to dream up some VSA representation that is essentially temporal, in which case it needs to be compatible with the time-scales of the implementation.
Are there better descriptions for the internal nodes? Something with intuitive meaning of their function would be good.
Does the integrated error needs to be initialised to zero? Does this imply the need for “turn-on” circuitry? Would it be better to have an “always-on” design? Should there be some error correction for cumulative errors?
There is a very minor technical question of whether it is better to represent the external signal sources and sinks by vertices in the data flow diagram or whether those data flows should be represented by edges with only one vertex.
3.2 Data flow tables
These table record more detail about the data flow diagram tthan can be reasonably displayed on the data flow graph.
Node and edge data flow tables go here
4 Design 02 - No VSA (tidied)
This is the previous design tidied slightly to be more compatible with the VSA changes which will be introduced later.
5 Extreme designs
Digress briefly to consider the extremes of where we might go with incoirporating VSA components into the design of the altitude hold circuit.
5.1 Minimal collapse
Replace all the edges in the data flow diagram with VSA equivalents.
Leave all the vertices in classical form.
The edges are equivalent to wires. A value is inserted at one end and retrieved at the other end. This would require a classical to VSA interface at the input of the edge and a VSA to classical interface at the output end.
This would seem to be an almost pointless thing to do. However, the classical values are assumed to absolutely accurate as mathematical idealisations. Different physical realisations will be inaccurate to some extent. A software implementation would likely represent the values as double-precision floating-point. A very constrained computer might represent the values with 8-bit integers. An analog computer might represent the values as voltages in some constrained range and with noise present in the signal.
Analogously, there can be range, distortion, and noise effects present in VSA representations. There can also be implementation noise, for example if VSA representations are implemented with simulated spiking neurons.
- The minimal collapse scenario would be useful for investigating the effects of VSA representation inaccuracy on the behaviour of the otherwise unaltered altitude hold circuit. This could indicate which values require high fidelity representations.
5.2 Maximal collapse
- Replace the entire set of vertices of the data flow diagram (with the exception of external sources and sinks) with a single vertex. That is, altitude hold is treated as a single, complicated function with no discernible internal data flows. For example, it might be possible to implement the altitude hold as effectively an interpolating lookup table.
Such a model might have advantages depending on the implementation technology.
6 Design 03 -
7 Design 04 -
8 Etc.
R version 4.1.0 (2021-05-18)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 21.04
Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.9.0
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.9.0
locale:
[1] LC_CTYPE=en_AU.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_AU.UTF-8 LC_COLLATE=en_AU.UTF-8
[5] LC_MONETARY=en_AU.UTF-8 LC_MESSAGES=en_AU.UTF-8
[7] LC_PAPER=en_AU.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_AU.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices datasets utils methods base
other attached packages:
[1] DT_0.18 DiagrammeR_1.0.6.1 readxl_1.3.1 here_1.0.1
loaded via a namespace (and not attached):
[1] Rcpp_1.0.6 pillar_1.6.1 compiler_4.1.0 cellranger_1.1.0
[5] later_1.2.0 RColorBrewer_1.1-2 git2r_0.28.0 workflowr_1.6.2
[9] tools_4.1.0 digest_0.6.27 jsonlite_1.7.2 evaluate_0.14
[13] lifecycle_1.0.0 tibble_3.1.2 pkgconfig_2.0.3 rlang_0.4.11
[17] crosstalk_1.1.1 yaml_2.2.1 xfun_0.24 stringr_1.4.0
[21] knitr_1.33 fs_1.5.0 vctrs_0.3.8 htmlwidgets_1.5.3
[25] rprojroot_2.0.2 glue_1.4.2 R6_2.5.0 fansi_0.5.0
[29] rmarkdown_2.9 bookdown_0.22 magrittr_2.0.1 whisker_0.4
[33] promises_1.2.0.1 ellipsis_0.3.2 htmltools_0.5.1.1 renv_0.13.2
[37] httpuv_1.6.1 utf8_1.2.1 stringi_1.6.2 visNetwork_2.0.9
[41] crayon_1.4.1