Reinforcement learning for adaptive cruise control

Hi,

I tried with the tutorial of applying DDPG method to adaptive cruise control(ACC) by https://www.mathworks.com/help/reinforcement-learning/ug/train-ddpg-agent-for-adaptive-cruise-control.html.

It seems working but when I check the pretrained model or after trainning my own model, the weights and bias in actorNetwork and criticNetwork are null, shown as follows:

How can I get the parameters? I am really confused.

Thanks,

Yixuan

Answer 1

[Emmanouil Tzorakoleftherakis]

Hello,

Try the following

load('SimulinkACCDDPG.mat','agent')  
actor = getActor(agent);
weights = getLearnableParameterValues(actor);

[Yixuan Wang]

Thanks!