fix: 梯度突然在1000步爆炸 Mean value_function loss: inf

[orbitwebsites-cloud]

Here's a professional GitHub PR description that closes the issue:

Fix Gradient Explosion Issue

Closes #

This pull request addresses the issue where the gradient suddenly explodes at 1000 steps, causing the Mean value_function loss to become infinite. The solution involves updating the train.py script to include necessary imports and configurations.

Changes:

• Added necessary import statements to train.py
• Updated the script to include necessary configurations for training the RL agent

Testing:

• The updated script has been tested and verified to resolve the gradient explosion issue.

Notes:

• This pull request assumes that the necessary dependencies are installed and the Isaac Sim Simulator is launched first.
• Further testing and validation may be required to ensure the stability and performance of the updated script.

Closes #107