This is a simulator that leverages pedestrian datasets to simulate pedestrians.

• Go to base directory

git clone https://github.com/allanwangliqian/dataset-crowd-simulation.git
cd dataset-crowd-simulation

• Set up a virtual environment

python3 -m venv .

• Install dependencies

pip install -r requirements.txt

• Set up RVO2. If there are any issues, please go to RVO2 for troubleshooting.

git clone https://github.com/sybrenstuvel/Python-RVO2.git
cd Python-RVO2
pip install -r requirements.txt
pip install Cython
python setup.py build
python setup.py install
cd ..

• Go to sim cd sim. Download the ETH, UCY and ATC(sample) datasets tar file from here and then extract.

tar -xvf datasets.tar

• Download the trained checkpoints for Conv3D Autoencoder from here, extract into the home folder. You should have a "checkpoints" folder in the project's base directory.

tar -xvf checkpoints.tar

• Make a data folder

mkdir data

• Use test_case_helper.py to check dataset pedestrian flows. Experiment with check regions and robot start and goal locations.
• Copy check_regions and start_end_pos to test_case_generation.py.
• Use test_case_generation.py to generate test cases.
• Initialize the simulator using the generated test files. See main.py for an example.