
Japan's KDDI and Michinori will test AI-driven optimization of mobile network quality for autonomous buses.
Starting September 1, the trial runs on two routes in Hitachi City.
The goal is to reduce network design labor and improve 5G connectivity for self-driving vehicles.
What happened
KDDI and Michinori Holdings announced on August 28, 2026, that they will run a demonstration from September 1 using AI to autonomously optimize communication quality for autonomous buses. The trial is part of a project adopted by Japan's Ministry of Internal Affairs and Communications and is conducted with cooperation from Hitachi City, Ibaraki Prefecture.
Why it matters
The demonstration compares network performance before and after AI-based RAN (radio access network) parameter optimization under remote monitoring. It aims to verify how much AI can cut the labor involved in network design and adjustment, and to measure improvements in 5G connectivity and signal quality for autonomous driving.
What to watch
The test will run on two routes: the Hitachi BRT route, which has operated level 4 autonomous driving over a 6.1 km section since February 2025, and the Omika Station loop route, about 3.0 km, which has run level 2 autonomous driving since January 2025. The BRT route includes 14 bus stops, 11 crossings with general roads, and 15 pedestrian crossing points.
Ask the AI about this article →
This trial is part of a broader Japanese government push to deploy level 4 autonomous driving in regional areas, as seen by its adoption under the ministry's project. The Hitachi BRT route already operates in revenue service at level 4, making it a realistic testing ground for network optimization. By comparing AI-adjusted RAN parameters against the status quo, the demonstration could show whether network tuning can be automated enough to cut manual labor—a key cost in running autonomous services. If successful, this could make it easier for other rural operators to adopt similar systems, though the results are not yet known. The inclusion of both a long BRT route and a shorter loop provides data from different operational scales and vehicle types, which may help generalize the findings.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
AI is now helping solve longstanding math problems, including the so-called Jacobian conjecture, and top mathe…

A benchmark tested 15 language models on two briefs requiring cited sources

GE Vernova raised its 2026 revenue guidance to between $45.5 billion and $46.5 billion from a prior range of $…

Dozens of protesters gathered Thursday outside 8VC, a venture capital firm in Austin, to protest Palantir's AI…

AMD introduced ROCm 10, the next version of its software stack for AMD Instinct GPUs, designed to streamline t…

ECRI, a nonprofit patient safety organization, is expanding its reporting system to include errors involving a…
