ROBOTIS-GIT / turtlebot3_machine_learning

Apache License 2.0
123 stars 82 forks source link

State size not getting updated - ROS2 Dashing #64

Closed matheuslara closed 1 year ago

matheuslara commented 1 year ago

I made the setup to train the machine model (ROS2 - Dashing or Foxy) and train the turtlebo3t: https://emanual.robotis.com/docs/en/platform/turtlebot3/machine_learning/#software-setup

But in episode 64 the simulation is always slow. Episode 64 is the moment when the algorithm starts training the model.

                 if global_step > self.update_target_model_start:
                     self.train_model(True)
                 elif global_step > self.train_start:
                     self.train_model()

I believe that as the state size is 4, it may be causing this slowdown.

When I change the state_size to another value, it says that the value sent by the environment will continue to be 4.

I tried changing the sample to 362, 360, 24 (simulation/gazebo/burger model file)... But nothing that changes the value 4 sent to the environment.

How do I change the state size of the environment? What do I need to do to stop this slowness when training the model?