MATLAB: Episode simulation doesn’t run while training DDPG

exampleobstaclesreinforcement learningReinforcement Learning ToolboxroboticsRobotics System Toolboxsimulationsimulink

I'm currently trying to run the "AvoidObstaclesUsingReinforcementLearningForMobileRobots" example.

The Reinforcement Learning Episode manager opens and updates when I set the exampel to train the agent (dotraining=true), however, the little robot on the binary occupancy map does not move.

When I halt the training, the robot will move until it hits a wall (I'm assuming its tracing the last episode's path from before the training was halted).

The robot will move with the pretrained agent. I've tried training agents in other ddpg examples too and not had any issues.

How do I make the robot move during the training?

Best Answer

Hi Alice,

This example has not been set up to update the visualization during training. If you follow add a MATLAB Function block following this example you should be able to update it.

Related Solutions

MATLAB: How to extract the trained actor network from the trained agent in Matlab environment? (Reinforcement Learning Toolbox)

You can collect the actor (or policy) from the trained agent with getActor. Then, you can use the actor to predict the best action from an observation wtih getAction.

% get actor representation
actor = getActor(agent);
% actor predicts an action given an observation
action = getAction(actor, observation)

If you want to go further, have a look at the Implement REINFORCE from scratch example

MATLAB: Is the DDPG episode rewards never change during the whole training process

Looks like the scale between Q0 and episode reward is very different. Try unchecking "Show Episode Q0" to see of the episode reward changes. I would then simplify the critic network to make sure it outputs values in a similar scale as the episode reward.

Best Answer

Related Solutions

MATLAB: How to extract the trained actor network from the trained agent in Matlab environment? (Reinforcement Learning Toolbox)

MATLAB: Is the DDPG episode rewards never change during the whole training process

Related Question