How do I visualize the environment simultaneously with the training of the ddpg agent?

2 次查看（过去 30 天）

Prashanth Chivkula 2020-9-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/594226-how-do-i-visualize-the-environment-simultaneously-with-the-training-of-the-ddpg-agent

评论： Prashanth Chivkula 2020-9-25

Sleigh.m

I followed the matlab custom rl template and example . I get the agent running , however I get a blank figure . I have attached the environment file for reference. Can Someone please help me with this?

3 个评论
显示 1更早的评论隐藏 1更早的评论

Jan de Priester 2020-9-17

Hi,

Consider the following approach: inside the plot function, create a figure and attach it to "this"

function plot(this)

% Initiate the visualization

this.h = figure;

% Update the visualization

envUpdatedCallback(this)

end

Then in the functino envUpdateCallback (at the bottom of your script), define how you want your plot to update with the change in environment. I worked with this toolbox to solve the mountain car problem. Here is the code I created to visualize the environment:

function envUpdatedCallback(this)

% Set the visualization figure as the current figure

figure(this.h);

clf

% Extract the cart position

X = this.State(1);

% Plot the cart

cartpoly = polyshape([0 0 0.0125*3 0.0125*3],[0 0.0125*2 0.0125*2 0]);

cartpoly = translate(cartpoly,[X 1/3*sin(3*X)]);

cartpoly = rotate(cartpoly,rad2deg(cos(3*X)),[X,1/3*sin(3*X)]);

plot(cartpoly,'FaceColor',[0.8500 0.3250 0.0980])

hold on