Artifacts of the ADS Bug-Fix Pattern Study

Introduction

This is the first bug-fix pattern study in the Autonomous Driving System (ADS) domain. Our proposed hierarchical classification of ADS bug-fix pattern study includes 15 syntactic and 27 semantic bug-fix patterns, as well as root causes, symptoms, modules, sub-modules, related algorithms, and detailed bug-fix actions. This taxonomy is instrumental for developers, testers, and researchers in developing automated bug detection and repair tools for ADS.

The published paper (currently unreleased) of this ADS Bug-Fix Pattern Study will be available at . Now you can access the pre-print version of this paper at

The artifacts of this project are also available at

Directory Structure

|--DIR_ROOT
    |--data
        |--bfp_label
            |--csv
            |--json
        |--bug-fix file-change dataset
            |--ApolloAuto_apollo
            |--autowarefoundation_autoware
            |--autowarefoundation_autoware.universe
            |--pr_files.json
    |--figs

Usage

• data/bfp_label contains the labeled data of pull requests from Apollo and Autoware GitHub repositories in two formats.
• data/bug-fix file-change dataset contains the bug-fix code changes of pull requests.
• figs contains the output figures generated by data analysis and data visualization.

Figure Analysis Examples

Heatmap

Pie

Bar

Comparison Bar

Stacked Bar

Paper Citation

This paper is accepted by FSE 2025, which is not published yet. Currently, we provide an arXiv version of this paper. We will update the paper citation once it is published.

If you use the data or code in this repository, please cite the following paper:

@article{chen2025comprehensive,
  author       = {Yuntianyi Chen and
                  Yuqi Huai and
                  Yirui He and
                  Shilong Li and
                  Changnam Hong and
                  Qi Alfred Chen and
                  Joshua Garcia},
  title        = {A Comprehensive Study of Bug-Fix Patterns in Autonomous Driving Systems},
  journal      = {CoRR},
  volume       = {abs/2502.01937},
  year         = {2025},
  url          = {https://doi.org/10.48550/arXiv.2502.01937},
  doi          = {10.48550/ARXIV.2502.01937},
  eprinttype    = {arXiv},
  eprint       = {2502.01937},
  timestamp    = {Mon, 10 Mar 2025 13:18:36 +0100},
  biburl       = {https://dblp.org/rec/journals/corr/abs-2502-01937.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}