Reinforcement Learning Toolbox train two agent

Question

hunson yang 2020-4-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/518073-reinforcement-learning-toolbox-train-two-agent

回答： Ari Biswas 2020-4-17

采纳的回答： Ari Biswas

I want to train a DDPG model like this architecture.

I train this model with 500 episodes and 1 episodes have 1000 step.

But when I run the m-file.

The result is train 'RL Agent_1' 500 episodes then train 'RL Agent_2' 500 episodes.

This result will let the parameters 'x1' in the environment cannot return to 'Subsystem_1'.

How to fix this problem?

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Ari Biswas 2020-4-17

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/518073-reinforcement-learning-toolbox-train-two-agent#answer_426648

Training or simulating a Simulink model with multiple RL Agent blocks is not supported at the moment. However it will soon be supported in the upcoming releases. Feel free to reply back with further information on what problem you are trying to solve.