TIER IV Releases Reference Design for Autonomous Driving System for Racing Karts

We have not been able to verify the primary source, so please refer to the original for details.

According to TIER IV, developers can work on developing and improving autonomous driving algorithms and AI models without building an entire system from scratch. This matters to research and educational institutions and software developers seeking to reduce the initial burden of entry. The fact that 240 teams, or about 600 people, used a shared platform also raises a practical question: how well can it function as a reusable public resource?

What is new about this announcement is that TIER IV has made its development platform, centered on Autoware, concrete for a relatively lightweight vehicle—a racing kart—and released it on GitHub. By presenting it as a reference implementation that includes not just software but also the vehicle, GNSS, IMU, an in-vehicle computer, a simulator, an online evaluation environment and a web platform, TIER IV is making the testing process available to outside developers as well. The contents of the release reflect the premise that autonomous driving development cannot be completed with code alone and requires moving between real vehicles and virtual environments. In a related article, we reported that TIER IV CEO Kato called for the “democratization of autonomous driving” through an open-source platform. This release can be seen as reinforcing that vision in practical terms. Unlike the corporate joint development exemplified by the MoU between TIER IV and Astemo, this initiative is designed to open up the entry point to development. In other words, it is closer to a phase of increasing participation by distributing a shared platform externally than to one of expanding individual collaborations with specific companies. It is consistent with TIER IV’s previously reported approach and appears to make its open-source-centered strategy more concrete. The broader industry implication is that the competitive landscape is expanding beyond vehicle performance to include supporting infrastructure such as simulation, evaluation environments, public repositories, and shared data and design information for training. The fact that 240 teams, or about 600 people, used the same foundation shows that reusable testing environments, not just the software itself, can shape both developer education and implementation speed. However, it is not clear how much of the release is standardized and where each participant’s own implementation begins. The focus will be on reproducibility in real vehicles, how frequently the evaluation environment is updated, how extensively it can be modified on Autoware, and whether it will continue to be used after the competition.

The publicly released platform, which includes TOM'S EV racing karts, GNSS, IMU, an in-vehicle computer, a simulator and a web platform, offers a direct reference for engineers in Japan who can work across vehicle control, positioning, in-vehicle computing and testing environments. Developing talent and conducting demonstrations are both issues for autonomous driving in Japan. A key point of interest is whether a shared platform like this can serve not only educational purposes but also as a foundation for improving development reproducibility.