Project that explores learning a motion planner and PID controller with time-varying gains from an existing policy (such as one learned by RL).
- Python3
- PyTorch: https://pytorch.org/
- OpenAI Gym: https://gym.openai.com/docs/
- MeshCat (optional, for animations): https://github.com/rdeits/meshcat-python
sudo pip3 install torch==1.2.0+cpu torchvision==0.4.0+cpu -f https://download.pytorch.org/whl/torch_stable.html
sudo pip3 install gym
sudo pip3 install meshcat
sudo pip3 install pyyaml
sudo pip3 install cvxpy
sudo pip3 install autograd
sudo pip3 install hnswlib
sudo pip3 install rowan
python3 examples/motionplanner.py --rl
python3 examples/motionplanner.py --il
python3 examples/motionplanner.py [--animate]