Metadata-Version: 2.1
Name: torchdriveenv
Version: 0.1.0
Summary: TorchDriveEnv is a lightweight 2D driving reinforcement learning environment, supported by a solid simulator and smart non-playable characters
Author-email: "Inverted AI Ltd." <info@inverted.ai>
Keywords: reinforcement learning,drive,RL environment,torch-drive-env,torchdriveenv,invertedai,inverted ai
Classifier: Programming Language :: Python :: 3
Requires-Python: >=3.8
Description-Content-Type: text/markdown
Requires-Dist: pandas
Requires-Dist: numpy
Requires-Dist: shapely
Requires-Dist: scipy
Requires-Dist: imageio
Requires-Dist: torch>=1.10.1
Requires-Dist: invertedai>=0.0.16
Requires-Dist: omegaconf
Requires-Dist: opencv-python
Requires-Dist: gymnasium
Requires-Dist: lanelet2
Requires-Dist: torchdrivesim>=0.2.1
Provides-Extra: baselines
Requires-Dist: stable-baselines3; extra == "baselines"
Requires-Dist: tensorboard; extra == "baselines"
Requires-Dist: wandb; extra == "baselines"
Requires-Dist: ipykernel; extra == "baselines"
Requires-Dist: moviepy; extra == "baselines"

# Installation

The basic installation of torchdriveenv uses an OpenCV renderer, which is slower but easy to install. PyTorch3D renderer can be faster, but it requires specific versions of CUDA and PyTorch, so it is best installed in Docker.

## Opencv rendering

To install the “torchdriveenv” with opencv rendering:
```
pip install torchdriveenv
```

To run examples:
Set the `$IAI_API_KEY` and `$WANDB_API_KEY`
```
pip install torchdriveenv[baselines]
cd examples
python rl_training.py
```

## Pytorch3d rendering

To install the “torchdriveenv” with Pytorch3d rendering:
```
docker build --target torchdriveenv-first-release -t torchdriveenv-first-release:latest .
```

To run examples:
Set the `$IAI_API_KEY` and `$WANDB_API_KEY`
```
cd examples
docker compose up rl-training
```
