Having low hit rate using Isaac Sim on Window

Hello, I have been trying to follow the Github repo below:

to train the Jetbot to avoid object and to reach the target cube. However, the hit rate is very low (less than 20%) when I run the train_d.py from the repo. I tried to modify the code and the hit rate is still very low, some in the repo suggested that might be a scaling issue and the hit rate might be higher if we use Liunx rather than Window.

May I know if there is some other way to solve the issue of low hit rate ?
Many thanks

Hi there, I’m not overly familiar with this repository, but common things to try would be to make sure that the observation values from sensors and states are correct and that the reward function can correctly guide the robot to completing the task.