The world's biggest electric-vehicle maker now has a humanoid robot with a confirmed look, sort of. On October 9, China's National Intellectual Property Administration published a design patent naming BYD as the holder of an "embodied intelligent robot," the first public glimpse of the design of BYD's first self-developed humanoid. The patent drawings show a bipedal machine with two arms, two legs, articulated fingers, and an oval head with a dark face, finished in a black-and-white color scheme with no display screen.

The design closely resembles "Xiao Di" (little D), a humanoid BYD executives previously teased on social media. But the company has not confirmed whether the patent corresponds to the final production model, and a design patent only protects appearance. As the EV-focused outlet CnEVPost noted, the filing establishes nothing about performance, progress toward mass production, or whether the eventual product will adopt this look at all.

What the patent actually shows

According to the patent description, the machine is intended for walking, maintaining balance, and interacting with its surroundings, with potential applications in service, education, and industrial settings. The drawings reveal joints at the shoulders, elbows, hips, and knees, with some mechanical structures exposed around the waist and legs. Light-colored panels cover parts of the torso and limbs, and the overall proportions are roughly human.

The previously disclosed "Xiao Di" figures give a sense of the intended scale: 1.61 meters tall, 58.5 kilograms, 31 degrees of freedom, and a dexterous hand with a repetitive positioning accuracy of plus or minus one millimeter. Those numbers, reported by CarNewsChina from earlier BYD disclosures, point to a machine aimed at fine manipulation tasks such as grasping objects, gesture interaction, and pointing-based explanations, not heavy factory labor.

A design patent is a snapshot of intent. It tells you what a company wants a robot to look like, not whether it can walk across a showroom.

The debut that did not happen

Enjoying this story?

Get the five most important stories in tech, every morning. Free.

There is a reason this reveal arrived as paperwork rather than a stage launch. On July 28, BYD confirmed to the China Securities Journal that it planned to unveil a humanoid robot in August at Di Space, its science and technology exhibition and experience venue. August came and went with no formal unveiling, and as of October 9 the product still has not been introduced.

This is worth dwelling on because humanoid robotics is an industry where the gap between announcement and reality has a body count. The field is crowded with impressive demos and thin deployment numbers, and BYD, for all its manufacturing muscle, has not yet shown its machine doing anything in public. A patent publication is the most minimal form of progress: cheap to file, easy to overread, and legally required to be published whether or not the product is anywhere near finished.

Why a carmaker wants humanoids

BYD's "Xiao Di": What Is Known

Specifications from prior BYD disclosures, as reported by automotive press. The October 9 design patent itself contains no performance figures.

Height
1.61 m (reported)
Weight
58.5 kg (reported)
Degrees of freedom
31 (reported)
Hand repeatability
±1 mm (reported)

All figures are approximate values from earlier company disclosures via press reporting, not from the patent itself. A design patent protects appearance only and establishes no performance, schedule, or mass-production status.

Still, dismissing BYD's robotics push as marketing would miss the strategic logic. BYD began building a dedicated robotics research team in 2022, and the company has been assembling a position in embodied AI for years. In August 2023 it invested in Agibot, a Chinese humanoid startup, and in April 2025 it took a stake in Percibot. It has also been testing other companies' machines in its own factories: Ubtech's Walker S1 industrial humanoid has been deployed at BYD's Changsha plant for material-handling and identification tasks.

The company has also telegraphed where it thinks the first commercial uses are. BYD executive vice president Stella Li said in June that she hoped to eventually deploy two to three robots at every BYD dealership to introduce and demonstrate vehicles to customers. Li sees humanoid robots for homes and the service sector as a huge market for BYD, though she acknowledged that robots cannot replace the emotional connection between human sales staff and customers. If robots eventually enter households, she added, BYD could sell them through its existing dealership network, extending automotive sales channels into a new business.

That last idea is the genuinely interesting part. Carmakers have spent a century perfecting the sales floor, and BYD's pitch is essentially that a humanoid is the natural evolution of the showroom. Li also said BYD plans to build an open robotics platform that accommodates both its own robots and products developed in collaboration with other companies, and that automotive AI and robotics share common technological foundations with manufacturing capability as the core strength.

The car-industry robot race is getting crowded

Humanoid robot design, rendered in a patent-style illustration
BYD's first in-house humanoid robot design, revealed through a CNIPA design patent published October 9, 2026. (Patent rendering: China National Intellectual Property Administration)

BYD is not alone in this bet. Xpeng has been pushing toward mass production of its IRON humanoid, with CEO He Xiaopeng taking direct charge of the robotics business, and Tesla continues to develop Optimus for factory work. The pattern is unmistakable: automakers see their factories, supply chains, and assembly expertise as a natural launchpad for physical AI.

There is real logic to it. Building cars at scale demands precision mechatronics, high-volume supply chains, and cost discipline, all of which transfer to humanoids. But the analogy has limits. Cars operate on roads built for them; humanoid robots are being asked to operate in environments built for humans, which is a far harder generalization problem. The car companies that succeed in robotics will be the ones that treat the software challenge, not the hardware, as the main event.

What to watch

Three questions will determine whether this patent matters. First, will BYD actually unveil the product, and will it do something live and unscripted when it does? Second, what happens with the dealership plan, since showrooms are controlled environments that could genuinely be the most realistic near-term market for humanoids. Third, how the open robotics platform takes shape, because that is where BYD could differentiate itself from rivals who are building closed ecosystems.

For now, what exists is a set of drawings and a set of numbers, both reported, neither demonstrated. That is normal for this stage of an industry, but it is also exactly why the sober reading of a design patent matters more than the excited one. The shape is here. The robot is not, quite yet.

References