Skip to content

Industry News

Japan Airlines Is Putting Humanoid Robots on the Tarmac at Haneda

Japan Airlines announced on April 27, 2026 that it will deploy Unitree G1 and UBTECH Walker E humanoid robots at Tokyo Haneda Airport starting May 2026, in Japan's first trial of humanoid robots for airport ground handling. The three-year experiment addresses severe labor shortages in Japan's aviation sector as inbound tourism hits record levels.

By Cara Voss · April 29, 2026

Japan Airlines Is Putting Humanoid Robots on the Tarmac at Haneda

Japan Airlines announced on April 27, 2026 that it will begin Japan's first humanoid robot trial in airport ground handling operations, deploying two Chinese-made robots at Tokyo Haneda Airport starting in May 2026. The three-year experiment will test whether humanoid robots can move cargo containers, open securing levers, and load baggage alongside human crews on the tarmac.

The trial involves two distinct platforms: the Unitree G1, a 132-centimeter robot weighing 35 kilograms, and the larger UBTECH Walker E, standing 172 centimeters and weighing 61 kilograms. Both robots run on NVIDIA compute and were selected specifically because their human-like form lets them operate in existing airport infrastructure without any structural modifications.

Key Stats

2

Robot Platforms Deployed

3 yrs

Trial Duration (to 2028)

60M+

Annual Haneda Passengers

May 2026

Trial Start Date

The Labor Crisis Behind the Trial

Japan's aviation sector is running into a structural wall. Inbound tourism set records in 2024 and 2025, pushing passenger volumes at Haneda and Narita to capacity. At the same time, Japan's working-age population has been shrinking for decades. Ground handling crews, who do physically demanding work in tight spaces around aircraft in all weather conditions, are among the hardest roles to fill and retain.

Ground handling is not like warehouse picking. Workers operate on active tarmacs, use a dozen different types of ground support equipment (GSE), and must meet strict safety certification requirements. Conventional fixed automation and single-function robots have struggled here because the environment changes constantly, equipment is not standardized, and the work requires the kind of reach-and-grip flexibility that has historically required a human body.

JAL Ground Service Co., Ltd. (JGS), the JAL subsidiary responsible for ground operations at major domestic airports, partnered with GMO AI & Robotics Trading Co., Ltd. (GMO AIR) to address this gap. GMO AIR, part of the GMO Internet Group, recently opened the GMO Humanoid Lab Shibuya Showcase on April 7, 2026, positioning itself as a commercial integrator of humanoid robot solutions in Japan. JGS brings 75 years of airport operations expertise; GMO AIR brings the robots and the software.

Why Humanoid Robots for Airports?

No infrastructure changes required: Human-shaped robots fit into existing airport layouts, jetways, and cargo bays without modification.

General-purpose adaptability: One robot can assist with baggage loading, operate cargo levers, and potentially handle cabin cleaning tasks.

Safety alongside humans: Humanoid form factors are designed for human-scale environments, reducing collision risk with equipment and personnel.

Labor shortage mitigation: Robots handle physically demanding repetitive tasks, reducing injury risk and workload for remaining staff.

Under the Hood: The Two Robots on the Tarmac

JAL and GMO AIR chose two platforms that represent opposite ends of the humanoid robot size spectrum. The combination is deliberate: different body sizes suit different tasks in airport ground handling, from tight cargo compartments to open tarmac cargo movement.

Unitree G1 (Compact / Kid-Sized)

The Unitree G1 stands 132 centimeters tall and folds to just 69 centimeters for storage or transport. At 35 kilograms, it is light enough for two people to lift and relocate quickly. The G1 carries up to 3 kilograms with dexterous three-finger grippers, moves at a top speed of 2 meters per second, and uses 360-degree LiDAR for perception. Compute is handled by an 8-core CPU or an optional NVIDIA Jetson module. The G1 retails in the range of 3,500 to 6,000, making it one of the most affordable full-humanoid platforms currently available. Unitree, the Hangzhou-based company behind the G1, has positioned the platform as a research and early commercial robot with an open SDK.

UBTECH Walker E (Full-Sized / Adult)

The UBTECH Walker E is the larger of the two. At 172 centimeters and 61 kilograms, it matches average adult human dimensions. Walker E runs on an NVIDIA Jetson AGX Orin AI module rated at 275 TOPS, giving it sufficient onboard compute for real-time vision and motion planning. It has 21 degrees of freedom and uses torque motors for smooth, force-sensitive joint motion. UBTECH, founded in Shenzhen in 2012, has been selling commercial humanoid robots since the Walker S series and has recently moved into automotive supply chain trials in China through a Honda Trading partnership.

Specification Unitree G1 UBTECH Walker E
Height 132 cm (folds to 69 cm) 172 cm
Weight ~35 kg ~61 kg
Degrees of Freedom 23–43 DOF 21 DOF
Payload Up to 3 kg Not specified
Top Speed 2 m/s Not specified
AI Compute 8-core CPU / NVIDIA Jetson (opt.) NVIDIA Jetson AGX Orin (275 TOPS)
Perception 360° LiDAR + cameras Vision + torque sensing
Approximate Price 3,500–6,000 Not publicly disclosed
Origin Unitree Robotics (Hangzhou, China) UBTECH Robotics (Shenzhen, China)
Commercial Status Available for sale globally Commercial pilot deployments

Why Chinese Robots at a Japanese Airline?

The selection of Unitree G1 and UBTECH Walker E reflects practical market reality: Chinese manufacturers currently lead on price-to-capability ratio for entry-level humanoid platforms. At 3,500 to 6,000, the Unitree G1 is accessible enough for multi-unit trials. Japanese humanoid programs, like Toyota's or SoftBank's legacy Pepper, target different market segments. For a commercial demonstration experiment focused on operational feasibility rather than long-term production commitment, Chinese platforms offer the shortest path from procurement to tarmac.

Who's Behind the Trial

JAL Ground Service (JGS)

Founded 1951. Handles ground operations at Japan's major domestic airports for the JAL Group, including aircraft towing and baggage loading. Brings safety certification expertise and operational domain knowledge to the partnership.

GMO AI & Robotics Trading (GMO AIR)

Trading arm of GMO Internet Group focused on commercial AI and robotics deployment. Opened the GMO Humanoid Lab Shibuya on April 7, 2026. Runs a "Humanoid Dispatch Service" and supplies and programs the robots for the JAL trial.

Unitree Robotics

Hangzhou-based manufacturer of the G1 humanoid. One of China's most prolific robotics exporters, with global sales via AliExpress and distributors. Filed a 10 million IPO on Shanghai's STAR Market in early 2026.

UBTECH Robotics

Shenzhen humanoid manufacturer founded in 2012. Currently in automotive supply chain trials with Honda Trading in China. Offered 124 million yuan (8 million) for a chief robotics scientist in 2026, signaling aggressive R&D scaling.

What This Means for Aviation and Beyond

The JAL trial is not just an airport story. Aviation ground handling is one of the most constrained labor environments in any industry: outdoor exposure, tight deadlines tied to flight schedules, safety-critical proximity to aircraft and fuel, and physical demands that cause high injury rates and turnover. If humanoid robots can prove useful here, the transfer case to ports, rail yards, and logistics hubs becomes much easier to make.

Japan is running this experiment for demographic reasons that are unusually clear-cut. The country's working-age population peaked in 1995 and has declined every year since. Tourism has grown rapidly in the opposite direction, reaching a record 36 million inbound visitors in 2024. The gap between labor supply and service demand is not a projection, it is already showing up as flight delays and reduced ground handling capacity at peak times.

36M

Japan inbound tourists in 2024 (record)

1995

Year Japan's working-age population peaked

75 yrs

JGS operational experience at Japanese airports

The selection of two platforms for a single trial is also notable from a procurement strategy perspective. Rather than committing to one vendor and one robot size, JAL and GMO AIR are running a parallel evaluation. That approach reflects how early the market still is: there is no obviously correct humanoid robot for tarmac logistics, and the trial architecture acknowledges that.

For Unitree and UBTECH, the JAL partnership is a marquee reference customer outside China. Both companies are competing for international ground-floor deployments before larger Western or Japanese manufacturers bring competitive platforms to market. A successful multi-year JAL trial, even in limited scope, becomes marketing collateral and technical proof point for every other airline and airport operator evaluating robotics.

The Bigger Signal

JAL is not a tech company experimenting at the margins. It is a 76-year-old national carrier with strict safety protocols and regulatory oversight putting humanoid robots on active tarmacs because the alternative, recruiting and retaining enough ground crew, is harder than integrating robots. When airlines start viewing humanoids as a workforce planning tool rather than a novelty, the deployment curve accelerates across the entire aviation sector.

What's Coming Next

The JAL trial is structured in phases over three years. The first phase, beginning May 2026, focuses on visualizing and analyzing airport site operations to identify where robots can be introduced safely. That means mapping workflows, identifying the specific tasks within reach of current robot capability, and establishing safety verification procedures with regulators. No specific performance benchmarks have been released for this phase.

Phase two moves into simulated airport environment testing before phase three attempts real-world task execution with full safety oversight. JGS has indicated interest in expanding the program to other domestic airports it serves if the Haneda trial produces usable results. Future tasks in scope include cabin cleaning and actual GSE operation, which would require higher dexterity and autonomous decision-making than cargo container transport.

May 2026: Trial begins at Haneda. Site analysis and initial robot deployment.

H2 2026: Simulated environment testing; first real-world task verification rounds expected.

2027–2028: Full phased verification through trial end; potential expansion to additional JGS-operated airports.

Post-trial: Decision on production deployment and additional robot procurement or platform changes.

Frequently Asked Questions

What specific tasks will the robots do at Haneda?

In the first phase of the trial, starting May 2026, the robots will focus on baggage and cargo handling on the tarmac. That includes transporting cargo containers, opening and closing securing levers, and loading and unloading luggage. Future phases aim to include cabin cleaning and Ground Support Equipment (GSE) operation, but those capabilities depend on how well the initial phase goes.

Why were Chinese-made robots selected instead of Japanese ones?

The Unitree G1 and UBTECH Walker E offer the best available combination of price, functionality, and commercial availability in 2026. The Unitree G1 costs between 3,500 and 6,000, which is low enough for multi-unit trials. Japanese humanoid programs are less commercially mature at this price point. GMO AIR, the commercial robot integrator in the partnership, selected these platforms based on capability match to the airport use case.

How long will the trial run?

The demonstration experiment is structured as a multi-year phased trial running from May 2026 through 2028. It proceeds in stages: site analysis and workflow visualization, then simulated environment testing, then real-world operational verification. JAL and GMO AIR have not set a fixed cutoff date for deciding whether to proceed to full deployment.

Are the robots replacing ground handling workers?

No, at least not in this trial phase. The stated goal is labor savings and workload reduction through human-robot collaboration, not headcount replacement. JAL frames the program as robots complementing human staff by handling the most physically demanding repetitive tasks, which should reduce injury risk and ease the burden on existing crews in a sector that is already understaffed.

Has any other airline done this before?

JAL describes this as Japan's first demonstration experiment for humanoid robots in airport operations. Other airports have tested autonomous mobile robots (AMRs) for specific tasks like baggage transport inside terminals, but humanoid robots working in active tarmac ground handling roles is a genuinely new category of deployment. The experiment is being watched by other carriers evaluating whether the approach scales.

The 12-Month Outlook

Japan Airlines' Haneda trial will generate real operational data on humanoid robots in one of the world's highest-stakes logistics environments. By year-end 2026, JGS should have enough field observations to determine whether the G1 and Walker E can reliably handle tarmac cargo tasks under live flight schedule pressure. That data will be closely watched not just by other airlines, but by port operators, rail logistics companies, and any industrial operator facing the same demographic labor constraints Japan is navigating now.

For Unitree and UBTECH, the JAL partnership provides a high-visibility international reference at a critical moment. Both companies are racing to establish global deployment credibility before the 2027-2028 window when US and European competitors expect to bring more capable platforms to market. A successful multi-year airport trial in Japan, covered by the BBC and Guardian on day one, is worth more than most marketing budgets.

The Bottom Line: Japan Airlines deploying humanoid robots on the Haneda tarmac is less a technology story than a labor economics story, and that makes it more significant, not less, because labor demographics do not reverse and airlines cannot wait for perfect robots.

The three-year timeline gives the trial room to fail early and iterate. If phase one finds that the robots cannot yet handle cargo securing levers reliably, the program can pause and resume when the software catches up. That measured, phased approach is exactly how serious industrial operators evaluate new technology, and it is the right way to read this announcement.