Skip to content

🤝 Contributing

Help us build a large-scale, ever-expanding benchmark! We highly encourage contributions via issues and pull requests, especially adding more navigation baselines.

🚀 Quick Start

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/amazing-feature
  3. Make your changes
  4. Run pre-commit checks: pre-commit run --all-files
  5. Commit and push
  6. Open a Pull Request using our PR template

✅ Pre-commit Hooks

The project uses pre-commit with Ruff for linting and formatting.

# Install hooks
./scripts/setup-pre-commit.sh

# Run manually
pre-commit run --all-files

# Format code
ruff check --fix
ruff format

Pre-push Hooks

Pre-push hooks run ruff check and ruff format --check before pushing. If blocked, fix the reported issues and try again.

📄 Pull Request Format

Use our PR template with the following title format:

[Priority Emoji] [Pattern]: Brief description
Emoji Meaning
🔥 New feature implementation
🚀 Urgent fix
🧪 Experimental / refactor
📝 Documentation / configuration
⚡ Cleanup / refactor

Examples:

  • :fire: Add: domain adaptation module with adversarial training
  • :rocket: Fix: urgent data collection script for demo
  • :memo: Docs: update README with installation instructions

PR Checklist

  • 🎯 Purpose: New feature / Bug fix / Docs / Refactoring
  • 🔁 Reproduce: Execution commands included
  • 📊 Changes: Summary of main changes
  • 🧪 Testing: Verified locally
  • 🤔 Review Focus: Areas where feedback is needed

📝 Documentation

We use MkDocs with Material for MkDocs.

# Install dependencies
pip install mkdocs-material pymdown-extensions

# Serve locally
mkdocs serve
Location Content
docs/ All documentation files
mkdocs.yml Documentation configuration

🎯 Roadmap

Completed

  • Paper release
  • Isaac Sim assets release
  • Nav2 support for rule-based navigation
  • Cost formula and reference sheet
  • Collected dataset with teleoperation
  • Imitation learning baselines

Coming Soon

  • Diverse maps (suburban, rural, port, orchard) beyond urban sidewalk
  • Additional robot platforms (different form factors, price points)
  • Various sensor configurations (stereo cameras, depth cameras, mmWave radar, ultrasonic)
  • Different hardware tiers (budget RGB-only vs. premium multi-sensor suites)
  • Expanded scenarios testing robustness under challenging conditions (rain, night, crowds)
  • Cost-aware reward shaping for RL training
  • Cloud inference cost modeling for VLA policies
  • On-device vs. cloud inference cost comparison
  • Open challenges and leaderboard for the community to beat baselines

🐛 Bug Reports

When reporting bugs, please include:

  • Operating system and version
  • Python version
  • Isaac Sim version
  • Complete error message / traceback
  • Steps to reproduce
  • Expected vs actual behavior

💡 Feature Requests

When suggesting features, please include:

  • Clear description of the feature
  • Problem it solves
  • Example use cases