Hey, I just tried it out and training in in progress.
However, I evaluate the checkpoint 8000.pt, robot just stands still, not responding to any velocity commands. Is this normal ? I am wondering how many iterations it takes to have a good performance?
And I also noticed that "sigma" of velocity tracking reward term is 5, which is very big...Do I need to fine tune any parameters to produce the results ?
Thanks for the reply!
Hey, I just tried it out and training in in progress.
However, I evaluate the checkpoint 8000.pt, robot just stands still, not responding to any velocity commands. Is this normal ? I am wondering how many iterations it takes to have a good performance?
And I also noticed that "sigma" of velocity tracking reward term is 5, which is very big...Do I need to fine tune any parameters to produce the results ?
Thanks for the reply!