Japan Airlines Deploys Humanoid Robots at Haneda Airport to Fight Ground Crew Shortage

The jet bridges, cargo bays, and tarmacs of Tokyo’s Haneda Airport were designed for human workers — which, it turns out, is exactly why they are about to become one of the most important proving grounds for humanoid robots in the world.

Japan Airlines Ground Service (JGS), the ground handling subsidiary of Japan Airlines, and GMO AI & Robotics Trading (GMO AIR) announced this week that they will launch Japan’s first trial deployment of humanoid robots at Haneda Airport beginning in May 2026. The experiment will run through 2028 and involve two robots: the Unitree G1, a 130-centimeter Chinese-built humanoid, and the UBTECH Walker E, a 172-centimeter model from another Chinese robotics manufacturer. Their initial mission is cargo transport and baggage loading — some of the most physically demanding and chronically understaffed tasks in commercial aviation.

The choice of humanoid form factor is not incidental. It is, according to the project partners, the entire point.

The Labor Crisis Driving the Experiment

Japanese aviation is facing a ground handling staffing crisis that has been building for decades and has now become acute. Japan’s working-age population has been shrinking since the 1990s, and the aviation sector — which requires workers available at all hours, in all weather conditions, performing physically demanding tasks for relatively modest wages — has been among the industries hit hardest.

The problem is not unique to Japan. Ground handling staffing shortages contributed to the record flight delays and baggage-handling failures seen at European airports in 2022 and 2023. Airlines in North America, Australia, and Southeast Asia have all reported similar constraints. But Japan’s demographic trajectory makes the challenge structurally more severe, and the country’s cultural ambivalence about large-scale immigration means that foreign labor is a limited solution.

Automation has been part of the answer for years. Baggage sorting in major airports is heavily automated, and passenger check-in has moved largely to kiosks and apps. But the most labor-intensive tasks — physically moving bags between carts and aircraft holds, loading cargo, conducting pre-flight checks — require work in unstructured environments that have resisted conventional robotic automation. Industrial robot arms and AGVs (autonomous guided vehicles) work well in controlled factory settings with standardized layouts. Airport tarmacs are full of obstacles, irregular surfaces, and unpredictable logistics flows that have defeated fixed-form automation.

This is where humanoid robots offer a theoretically compelling advantage. A robot built to human proportions can, in principle, operate in spaces designed for humans — reaching into the same overhead compartments, climbing the same access ladders, using the same carts and equipment — without requiring expensive infrastructure modifications. No new conveyor systems, no redesigned cargo holds, no special loading equipment. Just a robot that fits into the existing physical world.

The Robots: Unitree G1 and UBTECH Walker E

The Unitree G1, developed by the Hangzhou-based startup Unitree Robotics, has emerged in 2025 and 2026 as one of the most commercially accessible humanoid platforms. At 130 centimeters tall, 35 kilograms, and equipped with 43 degrees of freedom, the G1 can walk at up to 2 meters per second and features three-fingered dexterous grippers capable of handling a range of objects. Unitree has priced it starting at roughly $16,000 to $22,000 depending on configuration — cheap enough, the company argues, to make commercial deployment economics viable.

The G1’s affordability reflects a broader cost compression trend in Chinese robotics manufacturing. Unitree shipped over 5,500 units of its G1 and H1 series in 2025 and is targeting 10,000 to 20,000 units in 2026 — numbers that would have been unthinkable for a humanoid platform five years ago.

The UBTECH Walker E, at 172 centimeters, is closer to average adult human height and is designed for tasks requiring greater reach and strength. UBTECH, which is backed by Tencent and has been developing humanoid platforms since 2016, has positioned the Walker series for industrial and logistics applications. The Walker E’s physical dimensions allow it to operate in spaces dimensioned for adult humans without the cramped-posture compromises that shorter platforms sometimes require.

In the Haneda trial, the two robots will handle complementary roles during the cargo transport and baggage loading workflow. The specific task allocation — which functions the G1 takes versus the Walker E — will be refined as the trial progresses, with data collected on efficiency, safety incidents, and task completion rates.

GMO AIR and the “First Year of Humanoids”

The GMO Group company coordinating the trial, GMO AI & Robotics Trading, has positioned itself as Japan’s principal humanoid robot systems integrator. The company operates what it calls a Humanoid Dispatch Service — essentially a robotics staffing agency model where it supplies humanoid robots on a contract basis rather than requiring customers to purchase and maintain the hardware themselves. Earlier this year, GMO AIR opened a physical AI research hub in Shibuya designed to develop and test the motion programming and AI training pipelines required for commercial humanoid deployment.

GMO AIR’s leadership has publicly designated 2026 as the “First Year of Humanoids” in Japan — a framing that places the Haneda trial in the context of a broader national transition rather than a single corporate experiment. The Haneda trial is, in that framing, the first commercial proof point in a rollout that the company expects to expand to other airports, rail stations, and logistics facilities in Japan over the next several years.

JGS, which brings over seven decades of airport operations experience to the partnership, provides the operational domain expertise — the knowledge of exactly which tasks at exactly which points in the ground handling workflow are most suitable for robotic assistance, and which require human judgment and flexibility that current systems cannot replicate.

Context: A Week of Humanoid Milestones

The Haneda announcement lands in a week crowded with humanoid robot news. BMW began deploying humanoid robots in European automotive manufacturing facilities last month, marking the first large-scale humanoid deployment in Western industrial production. In April, Honor and the Monkey King team’s autonomous humanoid “Lightning” won the Beijing E-Town Half-Marathon in a record 50:26, demonstrating bipedal locomotion at speeds previously associated only with competitive athletes.

The cumulative effect is a rapidly compressing timeline between laboratory demonstration and commercial deployment. As recently as 2023, humanoid robots were primarily research curiosities. By 2026, they are loading luggage onto aircraft, assembling cars, and running half-marathons. The technology has not solved all of its hard problems — dexterous manipulation, robust operation in fully unstructured environments, and multi-hour battery life remain significant challenges — but it has crossed enough thresholds to attract commercial commitments.

The Haneda trial is notable precisely because it does not involve a cutting-edge custom prototype. It uses commercially available hardware from suppliers who are selling thousands of units per year. The innovation is in the deployment model — the partnership structure, the motion programming, the operational integration — not in the underlying platform.

What Success Looks Like

JGS and GMO AIR have not published specific performance targets for the trial, but the commercial case is straightforward: if humanoid robots can perform cargo transport and baggage loading at a fraction of the cost of hiring, training, and retaining human workers, the economics of deployment become compelling regardless of current capability gaps.

Japan Airlines is operating in an environment where ground handling staff turnover exceeds 20% annually at some airports, where peak season staffing is consistently insufficient, and where the demographic pipeline of young workers willing to take physically demanding airport jobs is shrinking. A robot that is available 24 hours a day, 365 days a year, does not require health insurance, pension contributions, or night-shift premiums, and can be redeployed between airports as demand shifts, addresses a very specific and very real cost problem.

The broader question — how quickly humanoid robots will move from carefully supervised cargo handling to the full range of ground operations — will depend on what the trial data shows. But the fact that Japan Airlines, one of the world’s oldest and most operationally conservative airlines, is running this experiment at its home airport signals that the age of the humanoid service worker is no longer a thought experiment. It has a start date.

It’s May 2026.