# Autoware — Open-Source Full-Stack Autonomous Driving Platform

> The leading open-source software platform for autonomous driving, providing a complete stack from perception and planning to control, built on ROS 2.

## Install

Save in your project root:

# Autoware — Open-Source Full-Stack Autonomous Driving Platform

## Quick Use
```bash
# Clone the Autoware meta-repository
git clone https://github.com/autowarefoundation/autoware.git
cd autoware

# Using Docker (recommended)
./setup-dev-env.sh docker
./docker/run.sh --devel

# Build inside container
colcon build --symlink-install --cmake-args -DCMAKE_BUILD_TYPE=Release
```

## Introduction
Autoware is the first open-source software platform for autonomous driving, maintained by the Autoware Foundation. It provides a complete self-driving stack covering sensing, perception, localization, planning, and vehicle control, all built on ROS 2. Autoware targets L4 autonomous driving and is used by research institutions, startups, and automotive companies worldwide.

## What Autoware Does
- Processes LiDAR, camera, and radar data for 3D object detection and tracking
- Performs high-precision localization using point cloud matching against HD maps
- Plans safe trajectories through complex traffic scenarios with behavioral and motion planners
- Controls vehicle steering, acceleration, and braking through a standardized vehicle interface
- Provides a simulation integration with CARLA, LGSVL, and AWSIM for testing

## Architecture Overview
Autoware is composed of modular ROS 2 packages organized into functional layers: sensing (drivers and preprocessing), perception (detection, tracking, prediction), localization (NDT matching, EKF fusion), planning (behavior, motion, parking), and control (trajectory following, vehicle commands). An API layer abstracts vehicle-specific hardware, enabling deployment on different platforms without modifying the core stack.

## Self-Hosting & Configuration
- Recommended setup uses Docker containers on Ubuntu 22.04 with NVIDIA GPU support
- Requires ROS 2 Humble or later as the middleware layer
- Build the full stack with colcon, the standard ROS 2 build tool
- Configure via launch files and parameter YAML files for each module
- HD maps in Lanelet2 format define road topology, traffic rules, and regulatory elements

## Key Features
- Complete L4 autonomous driving stack from raw sensor data to vehicle control
- Modular architecture allowing replacement of individual components
- Support for multiple sensor configurations and vehicle platforms
- Integration with open simulation environments for development and testing
- Active community with regular releases and global meetups coordinated by the Autoware Foundation

## Comparison with Similar Tools
- **Apollo (Baidu)** — comprehensive but tightly coupled to Baidu's ecosystem; Autoware is vendor-neutral and built on standard ROS 2
- **openpilot** — targets L2 driver assistance on consumer cars; Autoware aims at full L4 autonomy for robotaxis and shuttles
- **NVIDIA DRIVE** — commercial end-to-end platform; Autoware is free and open source with no licensing restrictions
- **Waymo Open Source** — provides datasets and tools but not a full driving stack; Autoware is a deployable complete system

## FAQ
**Q: What vehicles can run Autoware?**
A: Autoware supports any vehicle with a drive-by-wire interface. Reference platforms include Lexus RX, Toyota Prius, and various electric shuttles.

**Q: Can I test Autoware without a real vehicle?**
A: Yes. Autoware integrates with AWSIM (the recommended simulator), CARLA, and planning-only simulators for development and testing.

**Q: What sensors does Autoware require?**
A: A typical setup uses 3D LiDAR, cameras, GNSS/INS, and optionally radar. The minimum viable configuration requires at least one LiDAR and a GNSS unit.

**Q: How is Autoware governed?**
A: The Autoware Foundation, a non-profit organization, manages development with member companies contributing code, testing, and funding.

## Sources
- https://github.com/autowarefoundation/autoware
- https://autoware.org/

---
Source: https://tokrepo.com/en/workflows/asset-304857cb
Author: AI Open Source