Save listener Callback in eps format or any high resolution format

Hello,

for the Cart-Pole example and Training result (Episode Reward Chart) in RL Toolbox I am getting a static figure, that I can't save in anyway other than with the snipping tool which ruins the quality. I also tried to use the saveas command like here, but the eps or png file is coming empty and a figure 2 is opening.

Any help is much appreciated!

[Rik]

Have you tried using an explicit handle to the figure you want to store, instead of using gcf?

[Walter Roberson]

Are you using App Designer?

[Reinforcement Learning]

@Rik I am not sure, how I could do that. You mean something like this?

env = rlPredefinedEnv("CartPole-Discrete");
plot(env)

fig = gcf;
saveas(gcf, 'png_saved','png')

Figure(2) is opening and I am getting an empty file from that.

For the training there is a command rlPlotTrainingResults(trainingStats) but that still doesn't appear as a figure, so I don't know what exactly I should do. I opened the scripts of these commands, but I don't want to change something there.

[Reinforcement Learning]

@Walter Roberson No, just the Reinfrocement Learning Toolbox

[Rik]

You might need to use some tricks to get to the figure handle. The example below shows how you could try. I don't have the required toolboxes, so it is hard for me to experiment.

env = rlPredefinedEnv("CartPole-Discrete");
x=plot(env);
fig = getHiddenProp(x,'Figure');
saveas(fig, 'png_saved','png')

%confirm saving worked:
I=imread('png_saved.png');
figure,imshow(I)

function varargout=getHiddenProp(obj,PropType)
% Retrieve a hidden property from an object
% Use with care, this maight break for any release.
% If there are multiple objects, the first will be selected.
% If there are multiple matches, they will be returned in multiple outputs.
w=warning('off','all');
x=struct(obj);
warning(w);
if numel(x)>1,x=x(1);end
if numel(x)<1,varargout={};return,end
fn=fieldnames(x);
y=cellfun(@(elem) ...
    ismember(cellstr(PropType),...
    regexprep(class(x.(elem)),'.*\.','')... strip matlab.graphics.axis.Axes to Axes etc
    ),fn);
varargout=fn(y);
for n=1:numel(varargout)
    varargout{n}=x.(varargout{n});
end
end

[Reinforcement Learning]

@Rik I tried that and it worked. But the .eps file isn't really so clear. If I include that in my pdf, it's pixelated for some reason. Of course I changed the format from png to epsc

Answer 1

[Emmanouil Tzorakoleftherakis]

Hello,

If you are using R2020b, you can use

help rlPlotTrainingResults

to recreate the Episode manager plot and save it as you need.

If you are using R2021a, this function is now renamed to "inspectTrainingResult".

Hope that helps

[Reinforcement Learning]

Hi @Emmanouil Tzorakoleftherakis,

I am using R2020b. Do you mean like this? I already mentioned this in the comments, it only lets me save it as .pdf or .png. The bigger issue is with the Cart-Pole Visulaizer. It doesn't seem, like I can save that in any way