Summary

  • KDDI and KDDI SmartDrone said a 18 May test in Takahagi, Ibaraki, connected a flying drone directly to au Starlink Direct with Starlink's approval.
  • The aircraft exchanged control signals and telemetry while outside terrestrial mobile coverage.
  • KDDI calls it Japan's first drone flight controlled through direct communication with Starlink Mobile, as of 22 July.
  • The test did not demonstrate payload video, nationwide coverage, a commercial service or regulatory approval for beyond-visual-line-of-sight operations.
  • The next programme is meant to examine terrestrial-to-satellite handover and remote operation outside mobile coverage.

The useful part of KDDI's drone experiment is not that a satellite connected to another device. It is that the satellite link carried the small, consequential messages that keep an aircraft under control.

KDDI and KDDI SmartDrone disclosed on 22 July that they had flown the test on 18 May in Takahagi, Ibaraki. With Starlink's approval, the aircraft connected directly to au Starlink Direct in airspace where terrestrial mobile service was unavailable. Commands travelled to the aircraft and telemetry returned from it.

That is a more demanding operating role than an emergency text, even though the data volume may be modest. A control link has to remain timely and predictable enough for an operator to understand the aircraft's state. The demonstration therefore expands the practical surface of direct-to-device satellite connectivity without proving that the link is ready for every drone mission.

The control plane came first

KDDI did not say that the drone streamed inspection video through the satellite, moved a large payload dataset or maintained a guaranteed latency. The disclosed proof concerns command signals and dynamic aircraft information. Those are the control plane, not the full application workload.

That distinction matters for disaster response and infrastructure inspection. A drone may remain steerable while its camera feed uses another path, stores data locally or operates at a lower quality. Conversely, successful telemetry says little about whether a high-resolution live feed can survive the same link conditions.

KDDI describes the flight as Japan's first use of direct Starlink Mobile communication to control a drone. That is the companies' claim, measured as of the disclosure date. It is not an independent registry of every Japanese experiment.

Coverage is only one layer of the problem

Terrestrial mobile networks and the sky do not have identical coverage patterns. A route that appears covered on the ground can lose service at altitude or behind terrain. Satellite direct-to-device links offer another path, but a practical system must decide when to change paths and what happens during the transition.

KDDI says its next work, under a Japanese Ministry of Internal Affairs and Communications regional DX demonstration programme, will examine an optimal combination of mobile and au Starlink Direct links, including handover. It also plans to test remote operation beyond visual line of sight in areas without mobile coverage.

Those are future tests, not permissions already secured. Beyond-visual-line-of-sight operation brings aviation rules, command resilience, contingency procedures and human supervision into the design. A radio link alone does not settle them.

A national ambition still lacks operating evidence

The companies say they want drones to be remotely dispatched anywhere in Japan within ten minutes in normal times or emergencies. The May flight does not establish that network. KDDI disclosed no fleet size, dispatch locations, commercial tariff, customer, service-level commitment or nationwide availability date.

For operators evaluating the technology, the next evidence should be less theatrical and more operational: handover failure rates, command latency distributions, loss-of-link behaviour, power consumption, weather sensitivity and the separation between command traffic and payload traffic. Repeated flights across different terrain would matter more than another first.

The demonstration is nevertheless a real boundary crossing. Direct-to-device satellite connectivity was used to keep a drone's control loop alive where the terrestrial layer was absent. What it proved is precise: command and telemetry worked in one approved test. What remains is the harder work of turning that path into a governed, measurable service.

Sources