@inproceedings{imingail,
author = {Wei, Hua and Chen, chacha and Liu, Chang and Zheng, Guanjie and Li, Zhenhui},
title = {Learning to Simulate on Sparse Trajectory Data},
booktitle = {Joint European Conference on Machine Learning and Knowledge Discovery in Databases},
series = {ECML-PKDD '20},
year = {2020},
organization={Springer}
}
Usage and more information can be found below.
How to run the code:
The code relies on the simulator of CityFlow, which provides scalable, multi-process simulation for transportation sceanrios.
-
Intall the gym environment 'gym_citycar-v0' for CityFlow in your python environment.
https://github.com/wingsweihua/gym-cityflow-car.git
There are two branches, one for intersection network and one for ring network.
-
Pull the codes for ImInGAIL.
git clone https://github.com/wingsweihua/imingail.git
Start an experiment in the python environment with 'gym_citycar-v0' with the following script:
``bash runexp_gail_sparse.sh``
-
synthetic data
- simulation file: Traffic file and road networks can be found in
data/1_1
&&data/ring
. - logged data:
data/expert_trajs/1_1
&&data/expert_trajs/ring
- simulation file: Traffic file and road networks can be found in
-
real-world data
- simulation file: Traffic file and road networks can be found in
data/1x4_LA
&&data/4x4_gudang
. - logged data:
data/expert_trajs/1_4
&&data/expert_trajs/4_4
- simulation file: Traffic file and road networks can be found in