Visualize or plot Float32MultiArray data field

edit

This is due the the *MultiArray data types being quite ambiguous. As you have to / can specify the MultiArrayLayout, you can put rather arbitrary data in there. The docs give an example for encoding an image using this data type. So there is no straightforward way to implement a visualization of this.

Also, all the tools you mention actually work on single value data and plot respective time series. Even though you can configure rqt_multiplot a little bit more, I don't think you are able to do that with this kind of message easily.

You'd have to create your own specific plotting display, I guess.

PlotJuggler has this capability now with the lua scripting feature: https://github.com/facontidavide/Plot... though graph_rviz_plugin is much more straight-forward (you can add it to an .rviz file, launch that .rviz file from the command line or from a launch file and it will display though requires clicking the 'start' button on each plot, plotjuggler requires many clicks and selections to bring up a live display)

Duplicating much of the answer there:

Global

new_series = MutableScatterXY.new("trajectory_XY")

function

 new_series:clear()

  index = 0
  while(true) do
    str = string.format("/mobile_base/yaw_0/data.%d", index)
    series_yaw = TimeseriesView.find( str )
    if series_yaw == nil then break end  

    yaw = series_yaw:atTime(tracker_time)
    new_series:push_back(index, yaw)

    index = index + 1
  end

(Getting the hard-coded topic out of there is easy, turn this into one of the library functions with a topic as an argument:

function CreateSeriesFromMultiArray1D(new_series, topic, timestamp)
  -- new_series = MutableScatterXY.new(series_name)
  new_series:clear()

  index = 0
  while(true) do
    str = string.format("%s/data.%d", topic, index)
    series_y = TimeseriesView.find( str )
    if series_y == nil then break end    

    y = series_y:atTime(tracker_time)
    new_series:push_back(index, y)

    index = index + 1
  end
end

Global:

yaw_series = MutableScatterXY.new("yaw")

function:

CreateSeriesFromMultiArray1D(yaw_series, "/mobile_base/yaw_0", tracker_time)

But I don't see a one-step way to drag a multiarray onto the graph array and see the plot, some cut and paste and typing in various text boxes is required. )

It doesn't look hard to have additional rows of data that are plotted in parallel or become the x axis (the example library function shows that) or anything at all, so the flexibility is very high.

The scripts appear to be stuffed into ~/.config/PlotJuggler/PlotJuggler-3.ini as strings so the workflow of developing new scripts, checking them into source control and deploying them onto new systems looks awkward, and the scripts have to be worked on in the provided text box and not your text editor of choice (I'll follow up in the plotjuggler q&a about these).