Official implementation for the paper "Model-based Diffusion for Trajectory Optimization". Model-based diffusion (MBD) is a novel diffusion-based trajectory optimization framework that employs a dynamics model to run the reverse denoising process to generate high-quality trajectories.
Thank you so much for the feedback! I have updated the readme. For Car2D, since the environment is handwritten so it does not support Brax visualizer. If you want to visualize, PR are welcomed!
Thank you for open-sourcing the work!
I am trying to visualize the diffusion process by running the commands:
python visualize_mbd.py --env_name $ENV_NAME
However, I got the error of
AttributeError: 'Car2d' object has no attribute 'sys'
.I checked the car2d.py and it seems the sys is not defined in the Car2d class. Could you help with it?
Besides, I think the file for visualization has been renamed to 'vis_diffusion.py', you may also need to update it in the readme.