- Lane-Keeping with Hybrid Imitation and Reinforcement Learning (PPO)
- CNN for feature extraction via imitation learning and PPO for reinforcement learning.
- GPU relevant to RTX 3070 or better
- CARLA 0.9.15 Download for Windows
- Anaconda Distribution (conda 24.9.2 or later)
-
Clone the repository:
git clone https://github.com/daveshenal/Imitation-Initialized-PPO.git
-
Navigate to the project directory:
cd Imitation-Initialized-PPO -
Create PyTorch GPU Environment:
cd environments\pt_py38
conda create --name pt_py38 python=3.8 --file conda-requirements.txt
conda activate pt_py38
pip install --no-deps -r pip-requirements.txt
-
Create TensorFlow GPU Environment:
cd ..\tf_py310
conda create --name tf_py310 python=3.8 --file conda-requirements.txt
conda activate tf_py310
pip install --no-deps -r pip-requirements.txt
-
Activate the Environments:
conda activate <env_name>
- Author: Dave Perera
- Email: daveshenal281@gmail.com