''Error using horzcat '' when use Reinforcement Learning Toolbox to create a environment object for the Simulink model

Something wrong when I try to use Reinforcement Learning Toolbox to create a environment object for the Simulink model;

First I used the following syntax to load ActionObj and ObservationObj into the workspace

open_system('rlSimplePendulumModel')
env = rlPredefinedEnv('SimplePendulumModel-Discrete'); 
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Then I creat a DQN agent just using the syntax in the help documents of Reinforcement Learning Toolbox, then delete the object env in workshape, saved rlSimplePendulumModel.slx to current folder;

I really want to know how to create a environment object for the Simulink model finally when I executived this syntax

env = rlSimulinkEnv(rlSimplePendulumModel,[rlSimplePendulumModel '/RL Agent'],obsInfo,actInfo);

It returned a wrong statement saying : Error using horzcat. Dimensions of arrays being concatenated are not consistent.

I couldn't figure out what mistake I had made. Should I consider it as a bug caused by the Reinforcement Learning Toolbox?

I also used other predefined environment provided by the Toolbox and got the same mistake.

Answer 1

[Harsha Priya Daggubati]

Hi,

Can you try saving your model name in another variable and use it instead of the model name. I guess the error is with concatenation being done here.

[rlSimplePendulumModel '/RL Agent']

Try using it this way-

mdl = open_system('rlSimplePendulumModel');
env = rlSimulinkEnv(mdl,[mdl '/RL Agent'],obsInfo,actInfo);

Hope this helps!