The annual Autonomous Driving AI Challenge pits teams against each other using Autoware, open-source software for autonomous driving pioneered by TIER IV. Organized by the Society of Automotive Engineers of Japan (JSAE), the competition began in 2019, and TIER IV has supported operations since the inaugural event. This blog explores what competitors can learn from a technology provider's perspective and how those skills drive the evolution of the event.
The competition is growing rapidly, with the 2025 event attracting approximately 250 teams. The participant base is also becoming more diverse each year, welcoming competitors from a wide variety of backgrounds, ranging from students and working professionals to corporate engineers.
To support the expanding talent pool, the 2026 competition has been aligned with the SDV Skill Standards formulated by the JSAE in March 2025. This framework maps out the technical skills and career paths required across software domains in the software-defined vehicle (SDV) era, providing the industry with a common language for talent development. Applications for this year’s Autonomous Driving AI Challenge are open until July 31.
The 2026 competition features two distinct divisions designed to directly connect participants’ skills with roles required in the automotive and autonomous driving industries. The SDV Skill Standards define 31 roles that support SDV development, including in-car engineers, cloud engineers and support engineers. The AI Challenge divisions are structured so competitors can gain hands-on experience with the technologies required for several of these roles.
Based on Autoware and ROS 2, the Sim to Real SW Division combines rule-based algorithms and vehicle-to-everything (V2X) communication – a technology that enables vehicles to exchange data with one another and with roadside infrastructure – to improve driving performance. Competitors get the opportunity to deploy software developed in a simulation environment directly onto an actual vehicle.
The software for this division includes standard sample implementations from Autoware. These include the extended Kalman filter for traditional localization and prediction, Model Predictive Control for smooth predictive vehicle control and the Lanelet2 map format. All of these are core technologies used in actual autonomous driving.
Competitors can establish a baseline understanding of the software with the sample implementations. From there, they gradually progress to parameter tuning, algorithm replacement and building new features from scratch, ensuring participants experience the entire software development life cycle.
The exposure to V2X environments gives competitors the opportunity to explore the exact themes the industry is currently tackling for cooperative autonomous driving, such as receiving data from other vehicles via the cloud and integrating it into vehicle decision-making.
Participants can build skills spanning the core of SDV development, ranging from foundational software domains like middleware and software architecture to common automotive and function-specific domains like vehicle control, sensor processing and V2X.
V2X is a particularly broad domain. It goes beyond single-vehicle control and requires an understanding of cooperative control and communication standards. Working in this environment provides a rare practical opportunity where the skills of in-car engineers and cloud engineers intersect.
The other division in this year’s competition is End-to-End AI, in which participants will tackle a machine learning pipeline from the ground up. In this approach, a model directly outputs steering, accelerator and brake controls based on data collected by cameras and 2D LiDAR. The teams will build a system from scratch, starting from the data design phase.
Even in this division, standard sample implementations are provided to help participants get started. These include PilotNet, a convolutional neural network that directly outputs steering values from images, and TinyLidarNet, a lightweight model that directly outputs steering from 2D LiDAR point clouds. These technologies serve as accessible entry points into machine learning.
Teams will initially run their vehicles locally to get a feel for the system, before replacing the samples with custom models and refining data collection processes as they aim for higher rankings. The structure enables participants to experience the full cycle of data-driven engineering, including data collection planning, annotation design, model design, training, evaluation and testing.
This part of the competition begins with a data collection plan based on an operational design domain, moving beyond theory, all the way through to verification with cameras and LiDAR. With a strong focus on machine learning and data engineering, the division provides an opportunity to build the foundational skills of cloud engineers and domain specialists from a data-driven perspective.
During the competition, industry experts will host workshops on global navigation satellite systems, AI, V2X and real-world vehicle operations, reinforcing the learning experience for participants.
The AI Challenge is designed to run as a flywheel composed of three interlocking cycles.
It begins with the formation of a team and progresses to practical application in competitions, improving the technical skills of competitors. The cycle contributes to the Autoware community as high-quality code and insights are fed back into the ecosystem, driving the evolution of the open-source software and ultimately benefiting the industry as a whole.
The growing network of sponsors and partners provides the resources and technical expertise needed to develop new features in areas like AI and cooperative systems. This drives continuous improvements to the competition, which then attracts even more sponsors and partners.
Workshops, award ceremonies and promotion via social media raise awareness of the competition, expanding its reach and helping to increase the number of competitors and fans at subsequent events.
These three cycles all converge, elevating the value of the competition. Shared learning drives technological advancement, refining the competition platform and expanding the Autoware community as a result. The overarching goal of the competition is to build a structure where this mutually reinforcing cycle keeps turning.
No single organization can drive these cycles alone. It is only when the contributions of participants, sponsors, supporting organizations, research institutions and volunteer staff all come together that momentum is gained. The competition is structured to benefit everyone involved. Participants experience firsthand how their skills translate to the industry. Meanwhile, sponsors find an effective space to promote recruitment and technology, and research institutions gain a platform to validate hypotheses using real-world data.
The AI Challenge is both a competitive event and a flywheel that discovers, trains and steers autonomous driving engineers to the industry. By aligning with the SDV Skill Standards, the event is positioned to become a vital piece of the industry's talent pipeline. For participants, it is a place to test their technical skills. For companies, it is a venue to meet future engineers. For the wider industry, it is a platform for nurturing the next decade’s talent pool.
Registration deadline: July 31
Qualifying period: July 1 – September 1
Simulation final: September 19
Track final: September 20
Official website: en.jsae.or.jp/jaaic/index/
Taiki Tanaka | Autoware Scaling Department
Taiki joined TIER IV in 2020, initially working on autonomous driving systems for urban environments as a technical lead. Currently, he is a project manager overseeing operational support for the Autonomous Driving AI Challenge and also develops courses on building autonomous driving systems. Before TIER IV, he worked on AD/ADAS development at Nissan. He holds a master's degree from the University of Tokyo's Graduate School of Interdisciplinary Information Studies.
TIER IV is always on the lookout for passionate individuals to join our journey. If you share our vision of making autonomous driving accessible to all, get in touch.
We’re currently hiring for the following related positions:
Visit our careers page to view all job openings.
If you’re unsure which roles fit your experience, or if the current job openings don’t quite match your preferences, you can register your interest here. We’ll contact you when a suitable role becomes available and arrange an informal interview.
Inquiries
Social media
X (Japan/Global) | LinkedIn | Facebook | Instagram | YouTube
More