How to use the trained policy?

[nimesh00]

I trained the policy for widowGo1 environment and terminated it manually on saturation. I tried testing the trained policy using the script play.py but I'm confused how to command the robot. The robot just stands in place with the yellow cubes (that I'm assuming is the target ee region) moving around the robot but I cannot really understand how it is working. It just keeps moving along the red dotted line near the robot. I tried adding the following parameters to `play.py' but I did not see any change.

env.commands[:, 0] = 0.3
env.commands[:, 1] = 0.0
env.commands[:, 2] = 0.3
env.curr_ee_goal_sphere[:, 0] = 0.3
env.curr_ee_goal_sphere[:, 1] = 0.0
env.curr_ee_goal_sphere[:, 2] = 0.0

Can anyone guide me on what I'm doing wrong and how should I command the policy to move the end effector to a particular position?

[lyl1140292193]

I ’ve got the same question. I trained for 10000 iterations, and use "play.py" provided by "official leggedgym git". The dog just keep bouncing all the time, and the environment resets in a quick frequency.