Possibilities to use pointcloud as input for the SAC

Denys88 / rl_games

RL implementations

MIT License

800 stars 135 forks source link

Possibilities to use pointcloud as input for the SAC #249

Closed robinvetsch closed 9 months ago

robinvetsch commented 11 months ago

I use Isaac Gym including "rl_games" to train an RL-based path planning for a 6-DOF manipulator policy. The environment contains various obstacles that the robot must not touch.

Is it possible with the current release of rl_games to use point cloud data as input to the SAC agent, or do the 3D point clouds have to be flattened and passed to the MLP?

Many thanks for the help.

Denys88 commented 11 months ago

Hi @robinvetsch here is an example in readme how to create custom neural network. https://github.com/Denys88/rl_games/blob/master/rl_games/envs/test_network.py