2

I would like to know how I can add a trained RL controlled vehicle in the system in the tutorial01.

I have run the tutorial04_rllab. I added an RL controlled vehicle with the name “r1” and successfully trained it. Now, when I want to do the same in tutorial01 I cannot. It seems that this RL controlled vehicle cannot perform well.

Could it be because I didn't add the trained RL controlled vehicle?

Ashkan
  • 238
  • 1
  • 10

1 Answers1

1

When you train an RL policy in with rllab it generates a set of .pkl files that contain the contents of that policy, such as the weights of the network. You can visualize this policy by following the instructions here: https://flow.readthedocs.io/en/latest/visualizing.html. If you would like to run it in a new simulation as well, the weights/biases can be extracted from the pkl file as well; however, it is not something we currently support and you'd have to do some prodding around with visualizer_rllab. I hope this helps! Let us know if you have any other questions.

Fangyu Wu
  • 76
  • 6