Japan Airlines (JAL) will deploy humanoid robots in ground handling operations at Tokyo's Haneda Airport from May, in what is being described as Japan's first demonstration experiment of its kind at an airport. The two-year trial, conducted in partnership with GMO AI & Robotics, will initially focus on loading and unloading cargo containers, with plans to expand into cabin cleaning and ground support equipment operations. The trial is set to run in phases until 2028.

The robots in question stand 130 centimetres tall and are manufactured by Unitree, a company based in Hangzhou, China. At a media demonstration this week, one unit was seen pushing cargo onto a conveyor belt beside a JAL passenger aircraft before waving to a nearby worker — a moment that drew attention for its unexpectedly social quality. The machines can operate continuously for two to three hours before requiring a recharge. A key advantage of their humanoid design, JAL and GMO noted, is that they can be introduced without significant modifications to existing airport infrastructure or aircraft, unlike conventional fixed automation systems.

The trial comes as Japan's aviation industry faces mounting pressure from two converging trends: a surge in inbound tourism and a shrinking working-age population. More than seven million foreign visitors arrived in Japan in just the first two months of this year, following a record 42.7 million in the previous year. JAL currently employs around 4,000 ground handling staff, and industry estimates suggest Japan could need more than 6.5 million additional foreign workers by 2040 to sustain its economic growth targets. The government, however, is facing domestic political pressure to limit immigration.

Ground handling is physically demanding and requires skilled personnel — tasks such as aircraft marshalling and baggage loading impose considerable strain on workers, and conventional single-function robots have struggled to adapt to the complex, space-constrained environments around aircraft. Humanoid robots, proponents argue, offer the flexibility to navigate these workflows in ways earlier automation could not.