Connect with us

Robotic

Figure AI’s Helix 02 Gives the Figure 03 Full-Body Autonomy — and It Just Did Your Dishes

Published

on

Imagine walking into your kitchen and watching a humanoid robot — completely on its own — walk to the dishwasher, unload every dish, carry them across the room, stack them neatly in the cabinets, then return to reload the dishwasher and start it. No reset. No human hand-holding. Four uninterrupted minutes of real-world autonomy.

That’s exactly what Figure AI demonstrated on January 27, 2026, when it unveiled Helix 02 — the most capable AI model yet to control a humanoid robot’s entire body.

One Brain, Whole Body

The original Helix was already impressive: a single neural network controlling a humanoid’s upper body directly from camera pixels. Helix 02 blows that wide open. It extends unified neural control to the entire robot — walking, reaching, grasping, balancing — all as one seamless, continuous system.

What makes this remarkable is that it solves one of robotics’ longest-standing hard problems: loco-manipulation. Moving and manipulating objects at the same time has resisted clean engineering solutions for decades. The reason is deceptively simple — lift something and your balance shifts; step forward and your reach changes. Arms and legs constrain each other constantly.

Traditional robotics worked around this with clunky state machines: walk → stop → stabilize → reach → grasp → walk again. These hand-offs are slow, brittle, and deeply unnatural. Helix 02 ditches all of that.

The Architecture Behind the Magic: Systems 0, 1, and 2

Helix 02 runs on a three-layer hierarchy, each operating at its own timescale:

  • System 2 thinks slowly — interpreting scenes, understanding language, and sequencing multi-step goals.
  • System 1 thinks fast — translating perception into full-body joint targets at 200 Hz.
  • System 0 (new to Helix 02) executes at 1,000 Hz, handling balance, contact forces, and whole-body coordination simultaneously.

System 0 is the star. It’s a 10-million-parameter neural network trained on over 1,000 hours of human motion data and reinforcement learning across more than 200,000 simulated parallel environments. Rather than engineering separate reward functions for walking, turning, or reaching, System 0 learned to move the way humans move — stable, natural, and adaptive.

The result? System 0 replaced 109,504 lines of hand-engineered C++ code with a single learned prior. That’s not an incremental improvement — it’s a paradigm shift.

Figure 03: The Body That Makes It Possible

Helix 02 runs on the Figure 03 hardware platform, equipped with embedded tactile sensors and palm-mounted cameras. These unlock a new class of dexterity. With its “all sensors in, all actuators out” visuomotor policy, Figure 03 can now:

  • Extract individual pills from blister packs
  • Dispense precise syringe volumes in healthcare scenarios
  • Singulate small, irregular objects from cluttered piles — even when its own hands block the camera view

That last capability — handling self-occlusion — is a major milestone. Most robotic systems fail when they can’t see what they’re touching. Figure 03 feels its way through.

Why This Matters Beyond the Kitchen

The dishwasher demo is a compelling headline, but the implications stretch far beyond household chores. The same autonomy and dexterity that lets Figure 03 stack plates applies directly to manufacturing (handling irregular parts without pre-programmed fixtures), healthcare (precise medication handling with tactile feedback), and logistics (picking items of any shape in dynamic warehouse environments).

Figure AI has stated plans for 100,000 units of Figure 03, signaling this isn’t a research prototype — it’s a commercial play. As Helix continues to evolve, each software update makes every deployed robot smarter overnight, much like a smartphone receiving a new OS.

The Bigger Picture

What Figure AI is building with Helix 02 is something the robotics field has been chasing for decades: a robot that thinks and moves as a unified whole. Not a collection of subsystems duct-taped together, but a single learning system that perceives, reasons, and acts — continuously, in real time, in the real world.

The dishwasher task ran for four minutes without a single human intervention. That might not sound long, but in the timeline of humanoid robotics, it’s an eternity — and a preview of what’s coming next.

Humanoid Robots

Unitree Goes Public: China’s First Listed Humanoid Robot Maker Prices Its IPO at a $9 Billion Valuation

Hangzhou’s Unitree Robotics priced its Shanghai STAR Market IPO at a roughly $9 billion valuation — becoming China’s first publicly traded humanoid robot maker, with AI lab DeepSeek joining as a strategic investor and co-development partner.

Published

on

By

China just minted its first publicly traded humanoid robot company. On August 6, Hangzhou-based Unitree Robotics priced its Shanghai STAR Market IPO at 150.80 yuan per share, stamping the nine-year-old startup with a valuation of roughly $9 billion and kicking off one of the most closely watched listings of the embodied-AI era. Subscriptions open today, August 10, with the shares expected to begin trading between August 17 and 21.

For an industry that has run largely on venture capital, glossy demo videos, and promises about the future of physical labor, Unitree’s debut is a milestone of a different kind: a humanoid maker being priced by public markets, in real time, on its actual numbers.

A landmark listing for a nascent industry

Unitree is selling 40.45 million new shares — about 10% of its enlarged share capital — to raise roughly 6.1 billion yuan (around $850 million). The final price of 150.80 yuan came in about 45% above the market consensus of 104 yuan, a sign of just how much appetite there is for exposure to humanoid robotics. Investment banks expect a further re-rating once trading begins, with estimates ranging from 40 billion yuan to as high as 109 billion yuan in CCB International’s most optimistic scenario.

The company was founded in 2016 by engineer Wang Xingxing, who built Unitree around quadruped robots before pivoting hard into humanoids. Becoming China’s first listed humanoid robot maker gives Unitree not just a war chest but a public benchmark that rivals — foreign and domestic — will now be measured against.

The DeepSeek alliance: a brain for the body

The listing came with a headline-grabbing strategic partner. AI lab DeepSeek invested 140.8 million yuan (about $20.8 million) through a strategic placement, acquiring 933,399 shares — roughly 2.31% of the offering — under a 36-month lock-up. More important than the check is the accompanying co-development agreement: the two Hangzhou companies will jointly build AI models and embodied-intelligence systems for humanoid robots.

That pairing targets the central bottleneck in the field. Hardware has advanced faster than the software needed to run it, and the hardest problem in humanoid robotics remains the “brain” — a system that can interpret unfamiliar surroundings and translate a plain-language instruction into precise, safe physical action. Marrying DeepSeek’s large-language-model expertise to Unitree’s motion control and mechanical engineering is a bet that China can close that gap with domestic technology on both sides.

Why investors are betting on Unitree

Unlike many humanoid startups, Unitree is already shipping in volume — and making money. According to its prospectus, the company delivered more than 5,500 humanoids in 2025, with annual revenue quadrupling to 1.7 billion yuan and gross margins around 60%. That profitability stands in sharp contrast to some Western peers carrying enormous valuations on comparatively thin revenue.

The strategy is a full product ladder built around price. Unitree’s offerings include:

  • Go2 — its agile robot dog, starting around $1,600
  • G1 — a developer- and research-focused humanoid at roughly $13,500, widely cited as the cheapest genuinely capable humanoid on the market
  • H1 — a full-size humanoid listed near $90,000
  • H2 — its newest full-size platform priced around $29,900

By undercutting competitors on hardware and seeding developers with affordable machines, Unitree has positioned itself as the low-cost backbone of an emerging ecosystem — the company others build on top of.

What it signals for the humanoid race

A profitable, publicly listed humanoid champion — paired with a homegrown AI lab — reshapes the competitive map. It hands Unitree fresh capital and a currency for acquisitions and talent, and it gives investors a liquid way to bet on embodied AI for the first time. It also arrives amid tightening scrutiny of where connected robots are built and where their data flows, a backdrop that will shape how far Chinese humanoids travel into Western markets.

For now, the takeaway is simple: the humanoid robotics story is moving from pitch decks to public balance sheets. As the sector matures from demos into deployed, revenue-generating machines, Unitree’s listing may be remembered as the moment the market started pricing the robots that will share our factories, warehouses, and eventually our homes. At InteliDroid, we’ll be watching where the capital — and the machines — go next.

Continue Reading

Robotic

BYD Joins the Humanoid Race: Meet “Xiao Di,” the Robot Headed for Its Showrooms

BYD, the world’s largest EV maker, will unveil its first humanoid robot — nicknamed “Xiao Di” — in early August 2026, aiming to place service robots in dealership showrooms worldwide.

Published

on

By

The world’s largest electric-vehicle maker is about to add a new item to its lineup: a humanoid robot. BYD has confirmed that its first humanoid will make its public debut in early August 2026, and the company already has a specific job in mind for it — working the floor of its dealership showrooms.

The confirmation came after BYD released a teaser poster in late July hinting at an early-August reveal at its D Space (Di Space) facility in Zhengzhou. It’s the first time BYD has signaled a physical, operable prototype rather than just executive ambition, and it pushes one of the auto industry’s biggest names squarely into the humanoid robotics race.

Meet “Xiao Di,” a Robot Built for the Showroom

Early reports peg the robot’s nickname as “Xiao Di,” positioned as a service humanoid rather than an industrial one. According to specifications that circulated from a since-deleted BYD post — so treat these as unconfirmed until the launch — the robot stands about 1.61 meters tall, weighs roughly 58.5 kilograms, and carries 31 degrees of freedom across its body. The same leak claimed it can translate in real time between six Chinese dialects and six foreign languages.

Whether or not those exact figures hold, the design intent is clear. This is not a robot meant to lift sheet metal on an assembly line. It’s meant to greet people, hold a conversation, and sell cars.

Why a Carmaker Is Building a Greeter

BYD’s robotics ambitions have been telegraphed for months. Executive Vice President Li Ke said last month that the company hopes to deploy humanoids across its dealership network, where they could introduce vehicles to customers and assist with product demonstrations.

“My goal is to place two or three robots in every dealership,” Li said, describing machines that could explain vehicles, “create a more engaging atmosphere, and even demonstrate vehicle features.” He suggested robotic sales assistants could become commercially viable within the next one to two years, and framed home and service robots as a potentially significant growth market for the company.

It’s a shrewd first target. A showroom is a controlled, well-lit, predictable environment — far friendlier to today’s humanoids than a chaotic warehouse or a cluttered home. It also puts the robot directly in front of paying customers, turning a technology demo into a marketing asset for BYD’s core business of selling cars.

BYD Joins a Crowded Field

BYD is far from alone. A wave of Chinese automakers — including Xiaomi, XPENG, Chery, GAC, Changan, FAW, and Dongfeng — have launched humanoid programs, betting that their strengths in perception, control software, and mass manufacturing translate directly into embodied AI. On the global stage, Tesla continues to push its Optimus robot toward production, while startups like Figure AI have already logged nearly a year of real deployments at BMW’s Spartanburg plant.

What makes the automaker entrants worth watching isn’t any single robot. It’s the manufacturing muscle behind them. Companies that already build millions of vehicles a year — mastering supply chains, actuators, batteries, and cost engineering at scale — are precisely the players who could drive humanoid prices down fast if the technology matures. That’s the part of BYD’s announcement that should give pure-play robotics startups pause.

What to Watch at the Reveal

The August unveiling will be the first real test of how far BYD’s ambitions have progressed beyond slideware. The questions worth asking: Can “Xiao Di” actually hold a natural, multilingual conversation on a noisy showroom floor? How much of the demo is autonomous versus scripted? And is BYD building this in-house, or partnering with an established robotics supplier?

For now, BYD hasn’t officially confirmed the specifications, and a physical debut is still just days away. But the direction of travel is unmistakable. When the company that sells the most EVs on the planet decides its next showroom employee should be a robot, it’s a signal that humanoids are moving from research labs toward everyday commercial settings faster than many expected. We’ll be watching Zhengzhou closely — and we’ll report back once “Xiao Di” takes the stage.

Continue Reading

Humanoid Robots

URKL Debuts in Shenzhen: The World’s First Autonomous Humanoid Combat League

EngineAI’s new URKL league pits identical, fully autonomous T800 humanoids against each other in Shenzhen — a viral spectacle that doubles as a serious benchmark for balance, perception, and recovery.

Published

on

By

On July 16, a full-size humanoid robot took a spinning kick so hard its head flew clean off — and then it kept fighting. That surreal moment, broadcast live from Shenzhen, marked the debut of URKL, the world’s first professional combat league built entirely around autonomous humanoid robots.

A New Kind of Arena

URKL — short for Ultimate Robot Knock-out Legend — was founded in 2026 by Shenzhen humanoid maker EngineAI, and its opening gala turned a technical demo into prime-time spectacle. Aired on Guangdong TV with martial-arts star Donnie Yen in attendance, the event went viral when a robot called White Eagle kicked the head off an opponent named Matador, which carried on swinging regardless. Beneath the showmanship, though, sits a serious engineering benchmark.

Same Machine, Different Minds

Every competitor fights in an identical EngineAI T800 unit: 173 cm tall, 75–85 kg, with joints delivering up to 450 N·m of torque. Standardizing the hardware is the whole point. Because every robot is mechanically the same, victory comes down to software — perception, balance, decision-making, and motion control. Crucially, no human piloting is allowed. Once a bout begins, each robot fights entirely on its own AI, reacting to its opponent in real time. That single rule turns a boxing match into a live stress test of embodied intelligence.

The T800 itself is built to take punishment. It can throw uppercuts and spinning kicks, absorb hits, and — most impressively — pick itself back up quickly after a fall, a deceptively hard problem that has humbled humanoid developers for years.

Not Just a Knockout

URKL doesn’t crown winners on knockouts alone. Judges score robots across a spread of criteria, including:

  • Balance and stability under contact
  • Power systems and energy management
  • Motion control and agility
  • Real-time decision-making and perception
  • Structural durability and damage tolerance

In other words, a machine that stays upright, manages its actuators and battery intelligently, and makes good tactical choices can beat one that simply hits hard. It’s a scoring model that rewards the exact capabilities real-world humanoids need: resilience, autonomy, and graceful recovery from failure. The competition is already global — more than 200 teams registered worldwide, 32 advanced through online qualifiers, and the champion takes home a gold belt valued at 10 million yuan, roughly $1.44 million.

Why a Fight League Matters for Robotics

It’s tempting to write URKL off as gladiatorial theater, but combat is a shrewd testbed. Fighting compresses the hardest problems in humanoid robotics — dynamic balance, contact-rich interaction, fast perception, and recovery from catastrophic disturbance — into a few violent seconds. A robot that can take a kick, stumble, and stand back up has demonstrated exactly the robustness that factory floors, warehouses, and eventually homes will demand. Adversarial pressure also generates enormous volumes of edge-case data that scripted demos never produce. For EngineAI, one of a fast-growing cohort of Shenzhen humanoid firms, the league doubles as a recurring, televised proving ground for its hardware — and a marketing engine money can’t easily buy.

The Takeaway

URKL is equal parts sport, spectacle, and benchmark, and that combination is exactly why it fits the moment. As humanoid robots march from lab demos toward real deployment, the field needs public, repeatable, high-stress tests of what these machines can actually do when things go wrong. A headless robot that refuses to stop fighting is a viral clip today. It’s also a preview of the resilience the next generation of humanoids will need — and at InteliDroid, we’ll be watching to see whether the ring becomes the industry’s toughest test lab.

Continue Reading
Advertisement

Trending