Overview

The Autoware Centers of Excellence is happy to announce an incoming tutorial hosted at the IEEE Intelligent Vehicles Symposium in Cluj-Napoca Jeju Romania on June 22nd, 2025. This tutorial aims to introduce the open-source autonomous driving software Autoware Universe and to detail how it can be used in autonomous driving research. Autoware is an open source software platform and supporting ecosystem for autonomous driving research and development. Speakers are invited from universities and industry  worldwide to give talks on various autonomous vehicle related topics and the application of Autoware in these fields. The tutorial will serve as a general introduction to Autoware and the challenges in deploying autonomous driving systems in the real world. It will help the audience get familiar with installing Autoware, the technology underlying Autoware, and deploying Autoware on vehicle platforms and using Autoware in research projects. The tutorial will be split into 5 60-minute sessions, each covering a topic of interest. A detailed description for each session is provided below.

Organising Committee

Rahul Mangharam, University of Pennsylvania

Chi-Sheng Shih, National Taiwan University

Simon Thompson, TIER IV, Inc.

Audience

Professors, students, independent researchers, and industrial partners are welcome to attend. To fully enjoy its benefit, the audience should have a background in engineering, computer science, mathematics, robotics, or related fields. It is also helpful but not required to have an understanding of autonomous vehicles on either hardware or software.

Schedule

June 22nd, 2025 Full Day  9:00AM – 4:00PM EET  (all session times TBC)

8:50AM – 9:00AM  Introduction: Autoware as a research platform

9:00AM – 10:00AM  Session I: A Tale of Two Open-Source Ecosystems: Scaling Autonomy with AutoDRIVE and Autoware

10:00AM – 11:00AM Session II: Pix Platforms and Autoware Integration: Experience from Posnan University of Technology

11:00AM – 12:00PM Session III: Agnocast: Autoware Finally Achieves True Zero-Copy Publish/Subscribe IPC with a New Middleware

12:00PM – 1:00PM Lunch Break

1:00PM – 2:00PM Session IV: BeamNG Physics based Simulator and Autoware

2:00PM– 3:00PM Session V: CommonRoad Planning Benchmarks and Autoware

3:00pm – 4:00pm Session VI: AutoSeg – an open-source Vision Foundation model by Autoware Foundation

Sessions

Introduction

Simon Thompson
TIER IV, Inc

Welcome and Introduction to Autoware as a Research Platform

Session I: A Tale of Two Open-Source Ecosystems: Scaling Autonomy with AutoDRIVE and Autoware

Tanmay Samak and Chinmay Samak
Clemson University

Autonomous vehicle platforms of varying spatial scales are employed within the research and development spectrum based on space, safety, and monetary constraints. However, deploying and validating autonomy algorithms across varying operational scales presents challenges due to scale-specific dynamics, sensor integration complexities, computational constraints, regulatory considerations, environmental variability, interaction with other traffic participants, and scalability concerns. In such a milieu, we present AutoDRIVE Ecosystem, a unified framework for modeling and simulating digital twins of autonomous vehicle platforms across different scales and operational design domains (ODDs) to help support the streamlined development and validation of autonomy software stacks. Particularly, we discuss the adoption of autonomy-oriented digital twins to deploy the Autoware stack to achieve the ODD-specific objective(s) for various vehicles. Finally, we also discuss the flexibility of the proposed framework to support virtual, hybrid, as well as physical testing with seamless sim2real transfer.

Session II: Pix Platforms and Autoware Integration – Experience from Posnan University of Technology

Maciej Krupka
Posnan University of Technology

This tutorial is intended for first-time users of Autoware and provides a practical introduction to working with the PixKit/Pixloop platform. It will begin with an overview of Pixmoving’s ecosystem, followed by a step-by-step walkthrough led by members of the Autoware CoE community. The session will cover essential aspects of deploying Autoware on the vehicle, including workspace setup, installation, teleoperation, sensor calibration, and diagnostic tools offering a strong foundation for future development and research using this platform.

Session III: Agnocast: Autoware Finally Achieves True Zero-Copy Publish/Subscribe IPC with a New Middleware

Takahiro Ishikawa-Aso
TIER IV, Inc

Autoware has failed to introduce true zero-copy Inter Process Communication (IPC), despite the existence of the de-facto standard middleware: IceOryx. This is because the Autoware implementation heavily depends on unsized message types (e.g., those including std::vector as properties), although existing IPC techniques—including IceOryx—cannot achieve true zero-copy IPC for such ROS 2 message types. Throughout the codebase, Autoware uses unsized message types due to its design principles and its reliance on the ROS 2 ecosystem, where unsized message types are also widely used.

To address this problem, Agnocast, a new middleware designed to coexist with ROS 2, was developed. The Agnocast paper was accepted to ISORC 2025, and its production version has been integrated into Autoware. This tutorial introduces the principles behind Agnocast, details the integration process with Autoware, and explains how to use Agnocast in your own ROS 2 project.

Session IV: BeamNG Physics based Simulator and Autoware

Alexander Carballo
Gifu University

BeamNG is a physics based simulator used widely in racing car games. By modelling the physics of the vehicles and the environment, BeamNG allows realistic driving behaviour in various weather conditions and road surfaces. In this tutorial I report on our efforts to integrate the simulator with the Autoware autonomous driving stack, including the use of ROS2 bridge developed by our team at Gifu University.  We provide detailed instructions on how to set up the Simulator, the bridge and Autoware to facilitate realistic driving behaviour for Autonomous Driving research and development.. 

Session V: CR2AW: CommonRoad Planning Benchmark Framework and Autoware

Tobias Mascetta
Technical University of Munich

CR2AW extends Autoware by providing a rapid prototyping platform for motion planning researchers, which allows users to effortlessly transfer their algorithms from a simulation environment (CommonRoad) to a real vehicle (Autoware). Our code is open-source and we are currently running successfully on our TUM research vehicle EDGAR. CR2Autoware has already been used successfully by researchers for real-world experiments in multiple publications, including top-tier journals.

Session VI: AutoSeg – an open-source Vision Foundation model by Autoware Foundation

Zain Khawaja
Autoware Foundation

Autoware Foundation is proud to present AutoSeg, an open-source Vision Foundation model for autonomous driving. The AutoSeg Vision Foundation model has been designed and trained entirely in-house by Autoware Foundation exclusively using open-source datasets and we have publicly released all source code, model weights and training pipeline details. Our Vision Foundation model is highly efficient, enabling deployment on edge devices and covers a broad array of perceptual tasks essential for autonomous driving across multiple use-cases including depth estimation, general purpose scene understanding as well as end-to-end path prediction. We will present the details of our network architecture and training pipeline, as well as share insights on how we combined multiple open-source datasets into a unified schema to train our network. We will also describe how you can easily utilize AutoSeg in your autonomous driving applications and showcase some real-world examples where AutoSeg is being deployed by large industry players.