[Question] Deploying Stable Baselines 3 PPO-Trained Model on Jetson Nano for Real-Time Execution

[Gribesh]

This is related to gym-donkeycar. I cannot ask questions or issues in gym-donkeycar, so I am raising the issue here.

I have successfully trained a reinforcement learning model using Proximal Policy Optimization (PPO) through Stable Baselines 3 and the RL Baselines 3 Zoo frameworks. The model's performance has been validated within a simulation environment, and it works.

I am now at the stage where I wish to transition from simulation to deployment on a Jetson Nano for real-time operation. However, I am encountering uncertainty with the process of utilizing the manage.py file to initiate the model's runtime on the hardware.

My specific queries are as follows:

• What are the necessary steps to set up the trained model from Stable Baselines 3 on the Jetson Nano platform?
• Is there any specific configuration or dependencies that I should be aware of for the Jetson Nano to ensure compatibility with the model?
• How can I leverage the manage.py script, or is there an alternative method recommended for initializing and running the trained model on this device?

I would greatly appreciate guidance or references to documentation that may be helpful. Please let me know if any aspects of my question require clarification, and I will provide additional information.

Thanks.