← News

ASUS's white Gallop humanoid robot, with a visored head and two jointed arms with five-fingered hands, reaching toward a workstation beside a tray of small bagged motherboard accessories

13 Nov 2026

ASUS

ASUS's first humanoid robot, Gallop, is learning to pack motherboard boxes

ASUS detailed Gallop, its first humanoid robot, on Nov. 13, 2026: it is training at a real motherboard-accessory packing station in an ASUS group factory, learning mostly by watching the workers who do the job.

Most humanoid demos are dance videos and laundry folding. ASUS is doing the boring, correct thing: one real station, real parts in slippery little bags, and a pass/fail number it has to hit. It also owns the factories to test in, which most robot startups have to rent. The skeptic's note: 99.5% is a target, not a result, and the dexterous hand, the hardest part of any humanoid, isn't in the robot yet.

Gallop is ASUS's first humanoid robot. A humanoid is a robot with a body, arms, and hands built for the same stations a person uses. It comes from the Humanoid Robotics business unit of the Physical AI Solutions Business Group. Co-CEO S.Y. Hsu, one of the company's chief executives, introduced it on CNBC. The details are in a Nov. 13, 2026 post by Dr. Albert Chang. Those lines are ASUS's.

The first job is packing motherboard accessories at a station in an ASUS group factory. The plant is an EMS site, short for electronics manufacturing services, a factory that builds electronics. The station has 15 small parts. Many of them sit in thin clear bags. A person finishes a board in about 20 seconds. The target for Gallop is a 99.5 percent success rate inside that same 20 seconds, which Chang calls "the same standard we hold a person to." A 99.5 percent rate is about one miss in 200 boards. Those numbers are ASUS's, and they are a target.

ASUS says the full loop now works. A person shows the task, the robot learns it, and the robot is tested on the line. That claim is the company's.

Learning starts from leading open models, models published for others to build on. ASUS does not train its own giant model. Experienced workers wear small head cameras, with their consent, so Gallop can see how the job is really done. A simulation, a computer practice version of the station, then generates variations of that work. An operator in a virtual-reality headset and gloves guides the robot on the same parts and the same cameras. Once the robot is on the line, its recordings go back into training. Those steps are ASUS's.

The software is four layers. A planner breaks the job into steps. A second system, which ASUS calls a supervisor, checks every step and makes a new plan when something is off. A skills layer handles reaching, grasping, and placing. A control layer moves the arms and hands and stops at once if something is wrong. The layers meet at fixed interfaces, stable connections between them, so ASUS can swap in a better model without rebuilding the robot. Those lines are the company's.

One computer sits inside the robot, an NVIDIA Jetson AGX Thor. ASUS designs the body, and it designs the hand in-house, including a sense of touch. Motors and other standard parts come from partners. Those lines are ASUS's.

What is on the line today is a base that stays put, with two arms. Over the next year ASUS plans to put its own dexterous hand on the robot. Dexterous means able to handle small, fiddly parts. A wheeled version is planned too, plus a showing for customers at Computex 2027, the big computer trade show. Customer pilots, trial runs at a buyer's site, are planned for 2027. Sales are targeted for the second half of 2028. The program started in early 2026. Those dates are ASUS's.

Electronics manufacturing comes first. Hospitals and service sites come later, because a lot of the work is the same reach, grasp, and place. ASUS says Gallop is not meant to take people's jobs. It is aimed at repetitive stations that are hard to staff, and the people who teach it move into setting the robots up and supervising them. Those lines are the company's.

Chang's line in the post: "A demo shows that a robot can do something once. A factory needs a robot that does the job every day."

The picture is ASUS's white Gallop at the packing station, with a visored head and two jointed arms and five-fingered hands, reaching toward a workstation beside a tray of small bagged motherboard accessories. The credit is ASUS.

In plain terms, ASUS is teaching its first humanoid to pack 15 small motherboard parts, many of them in slippery clear bags, to the same 20-second pace and 99.5 percent bar it sets for a person. The company says the show, learn, and test loop is already running on a real line. The hand that can feel, the wheels, and a robot a customer can buy are still ahead: pilots in 2027, sales targeted for late 2028.

RELATED

ONLINE…

Comments

guidelines

Loading…

Loading…

Sources