How rtwotwo/autodrive works
This repository is a curated autonomous-driving research catalog and public evaluation hub. Visitors browse research papers across four tracks, explore public datasets, compare benchmark protocols, and inspect leaderboard entries. The catalog’s source data is split across JSON files; the README documents those capabilities. The tree and sampled scripts also show Zotero import and GitHub-star refresh workflows. The browser implementation and benchmark submission handling were not sampled, so their internal wiring is not inferred.
Main entry points: update_github_stars.mjs Star refresh · index.html Catalog page · sync_zotero.mjs Paper import · papers.json Paper tracks
Components 16
Web Catalog
- Catalog page
index.html - Browser interface
assets/app.js - Catalog styling
assets/styles.css
Research Catalog
- Paper tracks
data/papers.json - Public datasets
data/datasets.json
Evaluation Hub
- Benchmark catalog
data/benchmarks.json - Leaderboard results
data/leaderboard.json - Benchmark submissions
benchmark/submissions - Submission schema
benchmark/schemas/submission.schema.json
Data Curation
- Paper import
scripts/sync_zotero.mjs - Paper review
scripts/review_papers.mjs - Star refresh
scripts/update_github_stars.mjs - GitHub star cache
data/github-stars.json
Other components
- Research visitor
- Zotero export
- GitHub
Connections 11
- Research visitor → Catalog page: opens
- Catalog page → Browser interface: loads interface
- Catalog page → Catalog styling: uses styles
- Zotero export → Paper import: imports entries
- Paper import → Paper tracks: writes papers
- Paper import → GitHub star cache: reads cache
- Paper review → Paper tracks: updates metadata
- Paper review → GitHub star cache: updates cache
- Star refresh → Paper tracks: reads and updates
- Star refresh → GitHub star cache: reads and writes
- Star refresh → GitHub: fetches stars