Cite the paper

This work has been published in IEEE Transactions on Intelligent Transportation Systems. (https://ieeexplore.ieee.org/abstract/document/10359450)

Title: "Feasibility Evaluation of Oversize Load Transportation Using Conditional Rewarded Deep Q-Networks"

@article{son2023feasibility,
  title={Feasibility Evaluation of Oversize Load Transportation Using Conditional Rewarded Deep Q-Networks},
  author={Son, Hojoon and Kim, Jongkyu and Jung, Hyunjin and Lee, Minsu and Lee, Soo-Hong},
  journal={IEEE Transactions on Intelligent Transportation Systems},
  year={2023},
  publisher={IEEE}
}

Auto-Simulation for Swept Path Analysis

1. Docker build with shell script file

docker_build.sh

2. Docker run with shell script file

docker_run.sh

3. Start Automatic Labelling

cd train
python automatic_labelling.py

Binary Classification for Transportation Feasibility

cd train

Train & Test the classifier

python cnn_train.py

How to generate road segment data

1. Create json file that contains road shape parameters.

python draw/generate_roads_v2.py

2. Load roads from json file.

cd train
python load_roads.py

You can load original image and marked image.

How to use a manual labelling mode

1. Execute python code.

cd train
python automatic_labelling.py

2. Key Funcions

Q, W: Rotation(CCW, CW)

Arrows: Move up, down, left, right

T: Save as True

F: Save as False

R: Retry

C: Capture Current Pygame Image

Caution: Don't press the down arrow first