gym-like collision avoidance environment with ORCA
Tested on Ubuntu 18.04 and Python 3.6+
git clone [email protected]:navallo/collision_avoidance.git
cd collision_avoidance
pip install -e .
run ALAN just online learning
cd collision_avoidance/ALAN
python ALAN_true.py
run ALAN with action selction
cd collision_avoidance/ALAN
python Train_ALAN_action_space.py
-
OpenAI Gym (minimum install should be enough)
-
RLlib (Ray):
https://ray.readthedocs.io/en/latest/index.html
please install 0.7.0 dev from wheel link
https://ray.readthedocs.io/en/latest/installation.html#trying-snapshots-from-master
git clone [email protected]:navallo/collision_avoidance.git
cd collision_avoidance
pip install -e .
Training and testing:
user# cd collision_avoidance/
user# python run_rllib.py
just ORCA policy (not training):
user# cd collision_avoidance/collision_avoidance/envs/
user# python collision_avoidence_env.py
Tested with python3.7, Ubuntu 18