🚗 ApolloSimFuzz

Note

The documentation is currently incomplete. At this stage, the repository primarily provides a working pipeline for testing Baidu Apollo using a lightweight traffic simulator.

The framework has been tested to be stable on my own server setup.
Please feel free to join our Discord for quick discussions, and open issues for questions, bug reports, or suggestions.

Contributions are very welcome — whether it is improving documentation, fixing bugs, or extending functionality. I sincerely appreciate any feedback or support from the community.

ApolloSimFuzz integrates Baidu Apollo with TrafficSandbox, a lightweight traffic simulation framework, to support flexible, scalable, and closed-loop testing of Baidu Apollo’s decision-making functionalities in lightweight simulation environments.

The framework is designed for simulation-based testing and fuzzing, where Apollo interacts with a traffic simulator through perfect perception results and control commands, enabling systematic evaluation of decision-making robustness under diverse traffic scenarios.

---

✅ Pre-requisites

Before installation, please ensure the following dependencies are available on your system:

• Anaconda
• Docker
• NVIDIA Container Toolkit

⚠️ GPU support is required for running Apollo modules.

---

🛠️ Installation

Step 1: Navigate to the ApolloSim directory

cd Drivora/ApolloSimFuzz

Step 2: Run the installation script

bash install.sh

This script will automatically:

• Clone the official Baidu Apollo repository into ApolloSimFuzz/apollo
• Create a Conda environment named drivora-apollo-${APOLLO_VERSION}
• Install all required Python dependencies listed in requirements.txt
• Build the TrafficSandbox Docker image from ApolloSimFuzz/TrafficSandbox

📌 The Apollo version currently used is specified in the VERSION file.

---

⚙️ Build Baidu Apollo

Due to Apollo’s build and runtime requirements, several manual steps are required.

---

🔧 Step 1: Apply Apollo Patches

Navigate to the Apollo directory:

cd ApolloSimFuzz/apollo

Apply the patches located in:

ApolloSimFuzz/apollo_bridge/patches/

Move ApolloSimFuzz/apollo_bridge/patches/dev_start_ctn.sh to ApolloSimFuzz/apollo/scripts

Replace ApolloSimFuzz/apollo/WORKSPACE by ApolloSimFuzz/apollo_bridge/patches/WORKSPACE for building issues.

---

🏗️ Step 2: Build Apollo inside Docker

1. Start Apollo’s development Docker container

bash docker/scripts/dev_start.sh

You should see a success message indicating that the container has started correctly.

2. Enter the container

bash docker/scripts/dev_into.sh

3. Build Apollo

Inside the container, run:

./apollo.sh build

⏱️ The build process may take a significant amount of time depending on your hardware.

---

🚦 Build TrafficSandbox (Manually Install)

Step 1: Enter the TrafficSandbox directory

cd TrafficSandbox

Step 2: Build the TrafficSandbox Docker image

bash build.sh

After the build completes, verify the image:

docker images | grep drivora/sandbox

You should see the following image:

drivora/sandbox   latest

Step 3: Test

Check and modify config.yaml and then execute:

python start_fuzzer.py

---

📝 TODO

This project currently provides basic testing functionality (e.g., random testing) and is under active development.

• Provide comprehensive documentation and usage examples
• Refine minor components to improve robustness and stability

---

📬 Contact

For questions, contributions, or collaboration inquiries:

• Maintainer: Mingfei Cheng
• Email: snowbirds.mf@gmail.com
• Affiliation: School of Computing and Information Systems Singapore Management University

---

❤️ Sponsorship

If you find this project useful for research or development, consider supporting it via GitHub Sponsors.

📖 Citation

If you use Drivora / ApolloSimFuzz in your research, please cite the following papers:

@inproceedings{cheng2025decictor,
  title     = {Decictor: Towards Evaluating the Robustness of Decision-Making in Autonomous Driving Systems},
  author    = {Cheng, Mingfei and Xie, Xiaofei and Zhou, Yuan and Wang, Junjie and Meng, Guozhu and Yang, Kairui},
  booktitle = {Proceedings of the 47th IEEE/ACM International Conference on Software Engineering (ICSE)},
  pages     = {1--13},
  year      = {2025},
  organization = {IEEE}
}

@inproceedings{cheng2023behavexplor,
  title     = {Behavexplor: Behavior Diversity Guided Testing for Autonomous Driving Systems},
  author    = {Cheng, Mingfei and Zhou, Yuan and Xie, Xiaofei},
  booktitle = {Proceedings of the 32nd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA)},
  pages     = {488--500},
  year      = {2023}
}

📌 A consolidated .bib file will be provided in future releases.

---

📄 License

This project is released under the MIT License. See LICENSE for details.