czthehusky / ngsim_smarts

How to run

Step 0. Download the SMARTS submodule with git submodule init & git submodule update, make sure your SMARTS is at the right commit id (46c3497)

Step 1. Install SMARTS following the instructions in ./SMARTS folder

Step 2. Download the original data from https://jbox.sjtu.edu.cn/v/link/view/75a0931a222347e1ba2e0441407f4a1f and place it under ./ngsim folder

Step 3. Build NGSIM scenario with scl scenario build --clean ./ngsim (This will take a while)

Step 4. Generate expert demonstrations with python example_expert_generation.py

Step 5. Test rollout with python example_rollout.py (Alternatively, to accelerate sampling, you can check parallel_rollout.py).

Setup using Docker

If you have trouble with Step 1, you can configure SMARTS with docker.

1. Build the image

docker build . -t ngsim

2. Create the container

docker run -it -v $PWD:/src -p 8081:8081 --name <your_container_name> ngsim bash

Then you can continue from Step 2 inside the docker container.

Benchmark your policy

We provide a benchmark script for you to evaluate your model locally. Now the example script is just performing random actions policy, you can customize for your own policy models.

You can run the script with the following command:

python benchmark/benchmark.py <path_to_your_ngsim_scenario> --env_num 10