Newshot

Consumer embodied AI breaks the "dimensional wall"

AI summary

In the consumer embodied AI industry, there has long been debate over whether utility or companionship is the primary quality for robots accepted by mainstream consumers. Miaodong Technology's multi-terrain tracking robot Beni offers a new approach, arguing that utility and companionship are not an either-or choice. After raising over $5 million on Kickstarter and securing nearly 8,000 pre-orders, Beni became the highest-funded consumer robot in the platform's history and is set to begin global in-stock sales on November 10.

Translation(model translation, not the source's own text)

Article title: Consumer-grade embodied AI breaks down the "dimensional barrier"

Main text: In the consumer embodied AI industry, one topic has long been debated:

"For a robot that can truly be accepted by C-end users, which comes first—utility or companionship?"

Some say that for embodied AI to enter the home, it must be able to do work to truly create value.

So they pour more energy into specific scenarios—racing on models, data, and hardware—trying to make robots truly "useful" and a genuinely reliable "helper" in the home.

Ultimately, the value of companionship cannot be quantified—keeping a cat or a dog works just as well.

Others say that for embodied AI to enter daily life, providing emotional value is more important.

Cute at home, impressive when taken out, something to admire when friends visit, and a "chat companion" when lonely in idle moments—in life, a robot should be a warm and lovable "buddy."

If you insist on utility value, today's technical paths are not yet convergent enough, and even after all the competition, it would be hard to withstand the impact of big companies entering the market.

Who is right and who is wrong? Debated to this day, the topic still has no proven correct answer.

But perhaps many people have been thinking about it wrong—Miaodong Technology's multi-terrain tracking robot Beni offers a new idea: utility and companionship are not actually a "multiple-choice question."

This year on Kickstarter, Beni raised over $5 million, with nearly 8,000 units in pre-sales, becoming the consumer-grade robot with the highest crowdfunding amount in history. On November 10, the record-setting Beni will soon be available for immediate global sale, ready to face the commercial test of the massive C-end market.

Half is the "seawater" of practicality, half is the "flame" of companionship—to break down the "dimensional barrier" and combine both in a single product, what did Beni get right?

01

Useful, and not just "it can really shoot"

Practicality is key because "being able to do work" is the "entry ticket" for embodied robots to enter the home.

Such an autonomous tracking robot is an absolute "must-have" for many groups.

Many people have experienced scenarios like this:

What Beni solves is exactly this kind of "must-have."

As the world's first multi-terrain tracking robot, Beni's capabilities are actually reflected in two points: "imaging" and "embodiment."

Today, in the scenarios that embodied AI focuses on—whether industrial, retail, or household chores—very few can make the math between robots and human labor break even.

But imaging-type embodied products can truly turn "single shoots" that cost over ten thousand in service fees into "continuous ownership" priced in the thousands.

There are many companies that can do each of these two points well, but companies that can "grasp both" and turn the two into "one integrated game" are rare in the track—and Miaodong is one of them.

To serve the scenarios mentioned above well, being able to do imaging well and bring professional capabilities down into the embodied field is Beni's first principle.

First, born from a team with a DJI background, Beni's imaging foundation is solid:

4K 60fps ultra-clear footage, HDR color, 10-bit color depth, support for 2X zoom—compared with other flagship action cameras, its specs are no less impressive, maximizing "what you see is what you get."

But doing imaging well—specs are only the first hard threshold; also crucial to the final footage is the shooting angle.

The "ground-level" perspective that many worried about in the past turns out, in actual use, to be Beni's highlight. The little robot's lens stops roughly at the eye level of children and the line of sight of pets. When a dog runs past your feet, Beni can capture the image of its four paws alternately landing; flying grass clippings, a speeding skateboard, a tennis ball skimming the net, plus cleaner sky in low-angle shots—these details visible only close to the ground are Beni's home turf.

Consumer-grade embodied AI breaks down the "dimensional barrier"
Consumer-grade embodied AI breaks down the "dimensional barrier"

Having covered "shooting," now let's talk about "tracking." When shooting, Beni can keep up with people and also actively seek shooting angles—plenty of highlights:

Compared with rigid fixed-position phone shooting, Beni's camera movement is richer: rear tracking, parallel tracking, and 360° orbiting can all be achieved with automatic camera movement; even at high speed, the subject won't leave the frame.

If you want to shoot manually, on the companion phone App, users can also see Beni's real-time footage and achieve manual control.

With a usable camera in place, Beni's next problem to solve is stabilization—shake the lens while running and all the footage is ruined.

Miaodong's approach is a HybridFlow hybrid stabilization system that gives the footage double insurance: a mechanical gimbal stabilizes the lens, electronic stabilization preserves the image, taking the overall stabilization capability of shooting up another level.

For the footage shot, Beni can transmit it at high speed to a phone via Wi-Fi 6 and Bluetooth 5.4 for users to browse. For the highlights of the day, Beni's AI can automatically identify and generate a reel, editing the few seconds most worth keeping into a clip—one-tap finished video—so footage escapes the fate of "shot but never edited, gathering dust in storage."

Consumer-grade embodied AI breaks down the "dimensional barrier"
Consumer-grade embodied AI breaks down the "dimensional barrier"

For the footage shot, Beni is also quite strict about protecting user privacy:

Using embodied capabilities to lay the foundation for imaging

It shoots so well, but if you treat Beni as merely a simple "lens stand," then you underestimate it—as a multi-terrain tracking robot, Beni's embodied capabilities are not to be underestimated either.

Beni's core moat is a set of gimbal-body coordination logic that lets it maintain high-quality footage even during high-speed, complex movement. The robot's gimbal lens can read the chassis's real-time motion data, aligning the image with the robot's six-axis sensors to perform corresponding motion compensation—even when the robot moves, it can maintain image consistency, with better stabilization.

A simple example: if you just mount an action camera on the robot's head, many times you won't get the best footage—when the robot moves, the image should be compensated accordingly;

if the lens and the embodied body are disconnected and cannot read the robot's motion data, the shaking and jitter of the image will of course be more severe.

Unlike other home robots that can mount an external 360 camera, Beni has designed the stabilization and tracking logic at the底层 level, letting imaging capabilities and the robot body's capabilities join forces, so the final footage naturally comes out better.

With the underlying technical logic designed to guarantee the image, as an embodied robot with excellent performance, Beni runs fast—its top speed can reach 28.8 km/h. This speed does not sacrifice the body's robustness at the risk of wear, and can steadily keep up with running and cycling, and also chase after romping kids and dogs all the way while shooting.

Consumer-grade embodied AI breaks down the "dimensional barrier"
Consumer-grade embodied AI breaks down the "dimensional barrier"

Fast on flat ground, Beni is also very stable on uneven terrain.

Its dual wheel-legged structure can not only easily move and turn; when encountering steps, sidewalk edges, or steep grassy slopes, its 25.4 cm jump height lets it leap over in one bound, land steadily, and keep shooting.

Consumer-grade embodied AI breaks down the "dimensional barrier"
Consumer-grade embodied AI breaks down the "dimensional barrier"

For different terrains, Beni's wheels can also be easily switched—standard wheels handle everyday roads, off-road wheels handle grassy slopes and dirt paths, giving Beni a place whether on the road or in the park.

When encountering complex road conditions, Beni also has two wide-angle cameras on its chest for intelligent obstacle avoidance, enabling one-shot continuous recording across complex terrain; at a speed of 28.8 km/h, it can quickly identify obstacles and route around them.

Consumer-grade embodied AI breaks down the "dimensional barrier"
Consumer-grade embodied AI breaks down the "dimensional barrier"

If it really takes a tumble, the highly robust Beni is also tougher than it looks—it can get back up on its own and continue following, and with a set of custom protective shells, this "little guy" is quite durable and tough.

With so many features stacked together, this little robot weighs only 1.8 kg—the weight of a thin and light laptop—fits in a backpack, and is not subject to drone regulations. The quickly swappable battery supports 1.5 hours of endurance, and three batteries can run for four and a half hours. Leave battery anxiety behind—being usable at all times is what truly makes it "all-terrain."

Behind the little robot is a large accumulation of technology

Despite Beni's small size, truly uniting "imaging" and "embodiment" capabilities requires solid technical depth:

And what truly stops most teams is the "last mile" of mass production.

Miaodong's team told Leiphone.com: in the lab, making a prototype jump onto a step is not difficult; but in users' hands, making every one of tens of thousands of mass-produced units jump 25 centimeters, with consistent neural network behavior, and maintaining a longer structural lifespan—that is not simple.

Behind Beni's pair of big eyes is the fruit of a large amount of technology and engineering combined. But to truly break through the "dimensional barrier" between utility and companionship, Beni has another half—"a sense of life."

02

A sense of life, and not just "big cute eyes"

During one of Beni's field tests, the product team witnessed an all-too-"ordinary" scene.

On the road, a little girl just over 4 years old was having some friction with her mother nearby. The little girl was holding a plush toy, tears streaming down her face, and her mother's face was also dark—the atmosphere had almost dropped to freezing point.

At this moment, Beni came to their side, opened a pair of cute big eyes, gently stretched its legs, and performed a somersault for the two in their cold war—with this jump, the invisible barrier between mother and daughter was immediately broken, both couldn't help but smile, and the gloomy mood vanished.

Jumping up is the first little trick Beni shows after meeting many users. When children see Beni, their first reaction is to press its big head; when its head is petted, Beni will be "happy" enough to jump up—a capability the team calls "head-pat hop."

Consumer-grade embodied AI breaks down the "dimensional barrier"
Consumer-grade embodied AI breaks down the "dimensional barrier"

Not just children—the team told Leiphone.com that among all kinds of users of different ages, when Beni does a somersault or responds and interacts as if alive, it earns plenty of surprises and praise.

Article title: Consumer-grade embodied AI breaks down the "dimensional wall"

Body: This is just one facet of Beni's "sense of life"—but it shows that letting robots bring warmth and joy to users' lives is also a goal the team cares about deeply.

And many of Beni's design choices serve exactly this goal.

Anyone who has used Beni may feel the same: rather than a "robot," this "little guy" is more like the cats and dogs at home. When you walk past, it curiously fixes you with its big glowing eyes, tilting its head up just like a pet "begging for a scratch."

The product team put a lot of thought into those eyes: large LED beads are actually a rather clever choice, making Beni's eyes diffuse and its gaze softer, so being "watched" by it doesn't feel intrusive.

Within the team, the product manager who has both cats and dogs at home understands this design best—when you make eye contact with a pet, your body naturally releases a large amount of oxytocin, producing an overwhelming feeling of happiness.

So the team spent great effort polishing this "small matter" of eye contact, making everyday interaction warmer.

Eye contact with humans is still not the whole of Beni's "sense of life." To make this "silicon-based little guy" more like our companion, a wealth of clever touches were added, smoothing its everyday interactions with users.

In daily life, when Beni is knocked over, it sways and shakes its head as it gets up, like a cute little creature; during shooting, if an impact can't be avoided, Beni bravely picks itself up, prioritizing the protection of the shot above all.

Serious at work, free to act cute at other times—this contrast is exactly the surprise Beni's "sense of life" gives users.

To step into life's scenarios, Beni is not a "machine" waiting for the user to wake it up, but has behaviors of its own.

But too many behaviors annoy users; too few make it seem stiff. At home, Beni doesn't steal the show, doesn't disturb, yet still draws the user's attention to some degree. The Miaodong team calls this appropriate scale "weak companionship."

When home alone, pets have their own behaviors too—cats lick their fur, purr, and bask in the sun; dogs sleep and chase their tails. Beni's little daily behaviors—breathing, purring, and so on—are the unique source of its "sense of life."

When users want to play with Beni, it can also offer plenty of fun party games. Take coin hunting: once the game starts, it searches around for virtual coins, its sounds urgent as it nears a coin and bouncing happily after gobbling one up. Or tug-of-war: two users race their fingers, yanking Beni this way and that through the air, full of tension.

Many parents today go to almost any lengths to get their kids off the iPad—and Beni may be the "best playmate" to get kids outdoors.

Consumer-grade embodied AI breaks down the "dimensional wall"
Consumer-grade embodied AI breaks down the "dimensional wall"

Miaodong Technology says more new features and new interactive games will be delivered to Beni via OTA or firmware upgrades, and the SDK developer ecosystem will gradually open up, bringing users more fun and convenience and letting them experience a Beni that keeps growing.

Should embodied robots speak "human language"?

Beyond its appearance, Beni hums and purrs—the product team shared with Leiphone.com (WeChat official account: Leiphone) that after selecting AI-generated sound effects, they discovered Beni's voice was almost identical to their own Maine Coon.

And in the early design stage, among the many options the team considered, one that was temporarily set aside was having Beni communicate with users in human language.

This seems rather counterintuitive—many companion robots and emotional-companion AI hardware on the market today sell themselves on fun, thoughtful conversation.

Beni not speaking "human language" is actually a rather clever product design.

If it spoke human language, the last bit of unfamiliarity between Beni and humans would vanish, and users would unconsciously hold it to human standards, with more attention placed on verbal interaction.

To do verbal interaction, the industry's long-standing latency problem would have to be solved, and Beni's identity as a tracking camera robot would naturally be diluted.

Moreover, voice interaction has more degrees of freedom—and with more freedom, robots naturally make more mistakes. But once a robot makes a mistake, it feels "dumb."

The Miaodong team has always believed: a product's form should serve its scenario. Beni's gaze and hopping are enough to give users a real sense of warmth; doing human speech too early would inevitably be gilding the lily.

In the team's mind, rather than a silly "child," Beni is better off as a clever "kitten."

For a good product team, choosing what not to do may be more important than deciding what to do.

03

Conclusion

Many users who already have Beni have given it quite high praise. Leading overseas tech reviewer MKBHD (Marques Brownlee) generously gave Beni a perfect 10, the only one among more than ten new products reviewed in the same period.

Back to the question at the start: tool-ness and companionship—which one is primary?

Beni answers with a single machine—no need to choose.

Ranking first in the consumer embodied AI field by crowdfunding amount, with a delivered product experience that also pleasantly surprises. Beni's attempt to break down the "dimensional wall" between tool and companion has already shown that what the market wants is not an either-or answer.

And after two years of debate, what Beni truly punctures is a prejudice: being practical doesn't mean making cold tools; being cute doesn't mean being a useless showpiece. In the past, teams making tools disdained to talk about emotional value, and teams making companions were ashamed to admit "it has to be useful"—Beni welds these two things together: being able to work is the reason it gets taken home; having a sense of life is the reason it gets kept and loved.

And this answer may have been written from the very beginning in Beni's big eyes—letting everyone become the protagonist of their own life, rather than forever the recorder standing behind the camera.

Consumer-grade embodied AI breaks down the "dimensional wall"
Consumer-grade embodied AI breaks down the "dimensional wall"

Original article from Leiphone.com. Reproduction without authorization is prohibited. See reprint notice for details.

  • From crowdfunding hit to in-stock sale, Miaodong Technology's Beni robot pioneers imaging ...
  • Sequoia and Longzhu rush to invest, $4.6 million crowdfunded: Miaodong Technology's next hurdle is ...
  • Exclusive | DJI 10-year technical veteran leaves, quietly joins Miaodong Technology
  • Over $5 million crowdfunded, world's first all-terrain tracking robot Beni offline ...
  • Exclusive丨Prediction engine Prophetlab raises tens of millions in seed round, with joint investment from Ivy Capital and other institutions
  • Wondershare discusses the way at GDIC 2026: AI reshapes creation workflows, judgment remains creators' core competitiveness
  • Exclusive | Former Tencent senior researcher raises tens of millions, anchoring physical AI contact intelligence
  • From "knowing answers" to "able to execute": how Zoho brings AI into enterprise business
  • Exclusive丨Yuanhuan Intelligence raises hundreds of millions: using "meta-models" to give AI "intuition"
  • Alibaba Cloud ranks first in Hong Kong IaaS for six consecutive years, enterprise customer AI penetration reaches 95%
  • Xianyu appears at UN Climate Change Conference: AI accelerates idle item circulation, low carbon blends into daily life
  • Average age 25, post-00s young man leads team to top score in first Ant Lingbo embodied large model challenge
  • When charging is as fast as refueling, the boundaries of rental car road trips are widened
  • Qijing GT7 Ultra Comfort Edition pure electric/extended range officially launched! Down payment from 66,600 yuan, get an L3-level architecture shooting brake
  • Doubao Work exclusively launches "Workbench" feature, first covering 21 professional scenarios including finance, healthcare, and education

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by 雷锋网
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗