Wetour Robotics (WETO) launches Orchestra wearable AI brain platform
Rhea-AI Filing Summary
Wetour Robotics Limited furnishes a Form 6-K highlighting its Orchestra launch event, where it unveiled Orchestra as a wearable “portable external brain” that runs multiple AI models locally with no cloud calls.
The Orchestra hub is a small NVIDIA Jetson Orin-based edge device designed to coordinate sensors such as an sEMG wristband and chest camera with actuators like exoskeletons, drones, smart lights and AR glasses. Wetour is positioning Orchestra as a platform with an AI orchestration agent and a simple three-part protocol for third-party hardware, and announced a free Early Access program for startups and builders, including hardware, SDK, engineering and go-to-market support.
Positive
- None.
Negative
- None.
Key Figures
Key Terms
TOPS technical
Ultra-Wideband technical
sEMG technical
Jetson Orin Nano Super technical
Vision-Language-Action models technical
computational twin technical
UNITED STATES
SECURITIES AND EXCHANGE COMMISSION
WASHINGTON, D.C. 20549
FORM 6-K
REPORT OF FOREIGN PRIVATE ISSUER
PURSUANT TO RULE 13a-16 OR 15d-16 UNDER
THE SECURITIES EXCHANGE ACT OF 1934
For the month of May 2026
Commission File Number: 001-42536
Wetour Robotics Limited
(Translation of registrant’s name into English)
Room 7003
3300 N Interstate 35 Ste 700
Austin, TX 78705
(Address of principal executive offices)
Indicate by check mark whether the registrant files or will file annual reports under cover of Form 20-F or Form 40-F.
Form 20-F ☒ Form 40-F ☐
Incorporation by Reference
This report on Form 6-K (the “Report”) shall be deemed to be incorporated by reference into the registration statements on Form S-8 (File No. 333-291960 and Form F-3 (File Nos. 333-294373 and 333-295457) of the Company, including any prospectuses forming a part of such registration statements, and to be a part thereof from the date on which this Report is filed with the U.S. Securities and Exchange Commission (the “SEC”), to the extent not superseded by documents or reports subsequently filed or furnished.
Orchestra Launch Event
As previously announced on May 1, 2026, Wetour Robotics Limited (the “Company”) held its Orchestra launch event on May 28, 2026.
Statements made during the event regarding future plans, expectations and prospects, as well as any other statements that are not historical facts, may constitute “forward-looking statements” within the meaning of the Private Securities Litigation Reform Act of 1995. These forward-looking statements are based on the Company’s current expectations and assumptions and are subject to risks, uncertainties and other factors that could cause actual results to differ materially from those expressed or implied by such statements. For more information regarding factors that could cause actual results to differ materially, please refer to the Company’s filings with the U.S. Securities and Exchange Commission, which are available at www.sec.gov.
The presentation materials and speakers’ scripts used in connection with the event are furnished as exhibits to this Report on Form 6-K. A replay of the event and the presentation materials will also be available on the Company’s website at https://wetourrobotics.com.
EXHIBITS
| Exhibit No. | Description | |
| 99.1 | Orchestra Launch Event Presentation | |
| 99.2 | Speakers’ Scripts from the Orchestra Launch Event |
1
SIGNATURES
Pursuant to the requirements of the Securities Exchange Act of 1934, the registrant has duly caused this report to be signed on its behalf by the undersigned, thereunto duly authorized.
| Wetour Robotics Limited | ||
| By: | /s/ Nan Zheng | |
| Name: | Nan Zheng | |
| Title: | Chief Executive Officer | |
Date: May 28, 2026
2
Exhibit 99.1
Move, the world follows. WETO · Austin, TX · May 28, 2026
Everything you just saw — what's running behind it?
That's it. About the size of a box. One-piece resin shell.
67 T O P S Previous wearable AI: 2–6 TOPS We did 10× more Four models at once · zero cloud calls
Tonight, it ships.
is a portable external brain for the human body.
THE DEFINITION Three words. Each one literal. Portable On your body. Not in a server room, not in the cloud. It goes where you go. External Outside your skull. Your brain runs the body you were born with — this runs the body you grew into. Brain Not a metaphor. It takes signal from nerves, models the world, and sends commands back out.
THE ARCHITECTURE Just remember three words. Brain Orchestra where decisions happen Nerves sEMG Wristband · Wearable Camera reads you / what you see Limbs Exoskeleton · Drone · IoT Devices carry out the action
A conductor plays no instrument. They decide who plays, and when. That is
What's inside the box.
INSIDE · COMPUTE One module runs a whole stack of models Far more than four — here are four of them YOLOv8 30 fps object detection · VisionLink CNN + LSTM 8-ch EMG · detects motion start / stop CNN + Mamba continuous gesture recognition · <35ms 8B Language Model natural-language control NVIDIA Jetson Orin Super · all at once · zero cloud calls
INSIDE · BOARD We didn't just bolt together a dev kit NVIDIA stock dev kit 100 × 80 mm heatsink and fan built for the desktop Our carrier board Reduced 60% of the dimensions Automotive-grade PCB · swappable module I/O re-planned Every peripheral plugs straight in — no adapters.
INSIDE · BODY Battery · Cooling · Connectivity 5 / 8 / 16 hrs heavy / mixed / standby — the exo can recharge it Active cooling holds peak load without throttling, all day 4 tunnels Wi-Fi 6 · BLE · UWB · USB-C / pogo UWB: human position in 3D, 40 Hz, <10 cm — "point at the lamp" means that lamp.
There's a fifth thing on the card. Not a model — an AI Agent.
THE PROTOCOL Want in? We ask for three things. 1 Interface what the data looks like — channels, rate, units 2 Connection how it reaches us — BLE / Wi-Fi / USB 3 Semantics what the data means — heart rate, position, SpO2 Think of it as a skill.md — you write the three-line spec, and Orchestra's built-in AI agent (OpenClaw-style) handles the rest.
You bring the hardware. We bring the brain that knows what to do with it.
One more thing...
ORCHESTRA · GEN 2 Same brain. From a box, to a card. Gen 1 · Box → Gen 2 · Card
WHY IT SHRANK Same compute. We rebuilt three things. Battery smaller, denser Cooling fan removed, cooling into the chassis — silent Shell from resin to a single block of CNC aluminum Mechanical engineering doing AI engineering's job.
110 × 56 × 16.8 millimeters 175 g lighter than an iPhone Same 67 TOPS · models you enjoy · zero cloud COMING SOON
Back to tonight's star. Everything next — it can do today.
Conductor and VisionLink aren't two products — they're two reference nerves.
REFERENCE NERVE · 01 reads you within 33 milliseconds From your intent, to command fired. 8-ch sEMG · 9-axis IMU · high-accuracy UWB, fused two-layer classifier · 10-sec on-device calibration · no cloud
REFERENCE NERVE · 02 VisionLink — reads what you see Watches your path reads the terrain, the exoskeleton shifts gears Watches your back someone approaches, the wristband buzzes
The first portable platform to close the human ↔machine loop on a single edge brain.
THREE HORIZONS Three horizons Tonight · 2026 one brain, two nerves — already working in front of you 2–3 years others' nerves plug in, an open-protocol ecosystem Further out plug the human in as a compute node — and build a complete digital twin of each one
HORIZON 2 Five directions Conductor unlocks Eye tracking AR glasses Smart home Robotics data Drone swarms Same three-part protocol — the Agent handles the rest.
The data you accumulate becomes your computational twin.
THE COMPUTATIONAL TWIN It queries your twin 01 A humanoid in your kitchen to understand "put the plates away" 02 A drone swarm over a site to verify the command was yours 03 An actuator in another city that wants to do work for you You become an address — all the work flows to it.
We're not racing to build the VLA. We're racing to build the data factory it has to learn from.
CARBON ⇄SILICON Carbon and silicon Today the brain is on your body Tomorrow the brain on you and the brain in the cloud become one Further out your twin is how other intelligence finds you
In twenty years, silicon-bodied creatures will live in your home. Something has to sit in between, so they understand us.
The Orchestra is a beautiful artwork built from separate instruments
Move, the world follows.
Wetour Robotics M A Y 2 8 , 2 0 2 6 · A U S T I N You just saw it work. Now let's build it together. Nan Zheng · Chief Executive Officer
Wetour Robotics P A R T N E R S H I P What Orchestra does for your product. Three directions where we add real value to mature companies. EFFICIENCY Better workflows. USER EXPERIENCE Better products. CREATIVITY New things possible. F O U R W A Y S T O W O R K T O G E T H E R 01 Build on Orchestra 02 Build with us 03 Ship together 04 Create new categories
Wetour Robotics Industries we want to partner with. Mature companies across these verticals — let's talk. 01 AR / VR Headsets 02 Automotive Cockpits 03 Industrial Automation 04 Defense & Tactical 05 Smart Home Devices 06 Drones & UAV 07 Service Robotics 08 Humanoids 09 Prosthetics 10 Exoskeletons & Rehab 11 Physical AI Agents ... and more. Your category? If you don't see your category — that's probably the most interesting conversation we could have.
Wetour Robotics A N N O U N C I N G T O D A Y Orchestra Early Access For startups, builders, and the people making what's next. Free. No equity. No exclusivity. No catch. W H A T Y O U G E T H A R D W A R E Orchestra hub + VisionLink + Conductor S O F T W A R E Full SDK + engineering support L A U N C H Go-to-market help when you ship apply.wetourrobotics.com
Wetour Robotics C L O S I N G In the mobile era, you looked down at the screen. In this era, you look up at the world. L E T ' S T A L K info@wetourrobotics.com wetourrobotics.com Wetour Robotics @WETO_IR_TEAM
Exhibit 99.2
OPENING
Good evening everyone. I’m Lilian, CMO of Wetour Robotics. Thank you for being here.
There are many different people in this room tonight. Some of you build companies. Some of you build systems. Some of you invest in what comes next. Some of you are researchers, engineers, students, operators, writers, friends.
But tonight, I want to ask everyone to set those titles aside for a moment. Whatever you build, whatever you fund, whatever you study. Let’s come back to one identity we all share.
A person living through a real moment of change.
When something new starts to emerge, we usually try to understand it through what we already know. We ask: is this a new computer, a new phone, a robot, another app?
Every once in a while, something comes along that asks us to change the frame itself. Tonight is about one of those moments.
You are going to see something live tonight that we have never shown publicly before. But before we show it, I want to give you the frame for why it matters.
AI IS STILL WAITING INSIDE THE SCREEN
For the last few years, AI has learned to do something extraordinary.
It can read, write, reason, plan, and surprise us in ways that would have sounded ridiculous only a few years ago. But for most of us, AI still lives in one place. Inside a screen. You open a laptop. You type a prompt. You get an answer. Then you close the laptop, and the physical world around you stays exactly the same. That is the part that is beginning to change.
The next question is not only how smart AI can become. It’s what happens when intelligence enters the physical world?
And I don’t mean AI as another chat window. I mean intelligence that can understand the slope in front of you, the object you are looking at, the way your hand is preparing to move, and the action you are trying to take before you have to explain it.
That is where this next era starts to feel very different.
HARDWARE STAGING
Let’s look at how computing has moved over time.
Twenty years ago, this is what you needed to run a serious AI model. A room full of machines.
Then it fit in here. More capability than that entire room, and you could carry it.
Then here. A computer in your pocket.
Each generation brought computing closer to us. From rooms, to desks, to pockets. But closeness is not the same as understanding. This one is different.
It does not ask for a screen. It does not wait for a typed prompt. It listens in a different way. I’m not going to tell you what it is yet. That would be too easy.
I’m going to give it to Bin. He will reveal it later.
Hardware kept getting closer. But the deeper question has always been communication. How do we talk to machines? And how do machines understand us?
THE TRANSLATION BURDEN
Every interface we have ever used has asked us to translate ourselves. If you want to use a computer, you learn how it wants to be used.
You learn its interface, its buttons, the ctrl c ctrl v, and how to make your intention legible to the system. And if you want real control. You learn code.
That is an incredible thing humans do. We take something fluid, messy, intuitive, and internal, and compress it into something a machine can accept. We have done it for so long that it feels natural. But it is still translation. The machine has not really learned our language yet.
That is the reversal that matters.
THE BODY AS LANGUAGE
When people hear “human-machine interaction,” they usually think of an interface. A screen. A button. A menu. A prompt. A command.Something between you and the machine. But before words, before language, before even conscious thought, the body is already communicating.
When you reach for something, your muscles begin before you finish thinking. When your eyes land on an object, your attention is already visible. When your body leans, hesitates, turns, reaches, stops, or prepares to move, there is information there.
Not abstract information. Physical information.
The body is constantly expressing intent, often before we have words for it. The old interface asked the human to translate. The next interface begins when machines can listen to signals the human body already produces.
Not because technology needs to become more complicated. Actually, the opposite. The best technology usually feels like less technology. It meets people where they already are.
BIOLOGICAL INTELLIGENCE
There is another part of this that matters. Human beings are not just brains sitting inside bodies. The body itself is intelligent.
You pick up a glass of water. If the surface is slippery, your fingers adjust before you consciously decide to adjust them. You walk across uneven ground, and your balance changes faster than language could describe. You catch something falling, and your body moves before your mind has time to explain why.
That is biological intelligence. Millions of years of evolution built a physical reasoning system that works below conscious thought, in real time, in situations it has never encountered before. It is fast, embodied, contextual. Machine intelligence is powerful in a very different way. Neither is sufficient alone. A machine without human judgment fails at the novel. A human without machine coordination is limited to what one body can do.
One will not replace the other. The real power is in both, performing in concert.
HUMANICS
Most of the visible conversation in robotics today is about humanoids. And that matters. Humanoids ask an important question: What can a machine do instead of a human?
But there is another question we care about. What can a human become when machines work with them?
What new boundaries open up? What limitations dissolve? What does a person look like when their biological intelligence and a machine’s intelligence work as one system?
We call it Humanics. Human plus robotics. Human capability that does not stop where biology does.
That is the world we are building toward.
2
WHO WE ARE
We are Wetour Robotics. We are an Austin-based Physical AI company building around this new world we call Humanics. A world where AI does not remove the human from the loop. It makes the human more capable inside it.
And to build that future, there is one problem that has to be solved first. When sensors on the body, cameras in the environment, wearables on the person, physical devices in the room, and AI models are all trying to work together in real time, something has to coordinate the relationship between them. Something has to understand intent, context, and action together. That is the problem our team has been building toward. We call it Orchestra.
Tonight, we are not going to explain all of it with slides. We are going to show you what it can make possible.
DEMO TRANSITION
The next part is where this becomes visible.
First, you will see a short film: a person moving through real space with our perception system and an exoskeleton working around them. Then Bin will come on stage for the live demo, where a human signal becomes real-time coordination across the room.
Different forms, same direction: a new way for humans and machines to work together through the body.
Let’s begin.
[Vision Link video begins]
Chief Technology Officer – Lian Bin
SLIDE 1 — Title: ORCHESTRA wordmark
What you just watched in the video — the chest camera detecting the person walking up behind you, the exoskeleton speed shifting on a finger count, the dashboard rendering eight channels of muscle signal at thirty hertz —that was VisionLink.
What you just watched me do on this stage — pointing, snapping, locking devices, flying the Tello, handing one of you a wristband — that was Conductor.
Both of them are nerve systems. Conductor is the nerve that reads me. VisionLink is the nerve that reads what I see.
SLIDE 2 — Hero shot of Orchestra Gen 2 card
Every signal that just fired, every prediction that just rendered on the screen behind me, every command that just hit a device in this room — all of them live here.
This is the Orchestra hub. One hundred ten by fifty-six by sixteen-point-eight millimeters. One hundred seventy-five grams. CNC-machined aluminum unibody. Sixty-seven TOPS of NVIDIA edge compute. Sitting in a shirt pocket. Right now sitting against my left rib, behind the fabric, with nothing tethered to it.
SLIDE 3 — Diagram
It is what the rest of the system is plugged into. The wristband — that’s a sensor. The chest camera on this exoskeleton — that’s a sensor. The exoskeleton itself — that’s an effector. The drone is an effector. The smart lights are effectors.
And these — the AR glasses I’m wearing tonight — are still a research prototype. I’ll come back to them later. For now, just notice that they are here, and the dashboard I’m reading off of them is coming from the same card.
The card is not a hub. It is a brain. The sensors are not inputs — they are nerves. The exoskeleton, the lights, the drone are not outputs — they are actuators. Hold those three words. I will use them for the rest of this talk.
I’m going to spend the next eighteen minutes on what is inside this brain, what we had to build to get it here, and what it does next.
3
SLIDE 4 — One-line definition centered on dark slide
Let me define it cleanly. Orchestra is a portable external brain for the human body. Not a hub. Not a phone replacement. Not a wearable companion. A second brain — outside the skull, wearable, computationally honest about what brains do. Three words in that sentence are doing work.
Portable — it lives on the body. Not in a server room. Not in a backpack tethered to a laptop. Not in a cloud region you have to trust. It is sized to be carried by a person who has forgotten they are carrying it.
External — outside your skull. Your biological brain is busy with the body it grew up in. This one is dedicated to the body you grew into — the exoskeleton, the wristband, the camera, the eyes that aren’t yours, the limbs that aren’t yours yet.
Brain — not in the metaphorical sense. In the structural one. It does what brains do. It takes input from afferent nerves. It builds a model of the world it lives in. It sends motor commands out through efferent ones.
The two nerves you have already seen — Conductor and VisionLink — are afferent. They carry signal in. The exoskeleton, the lights, the drone, the speaker — those are actuators. They carry action out. The card in between is the brain.
SLIDE 5 — Why “Orchestra”: conductor + players metaphor
We chose the name Orchestra on purpose. In a symphony, the conductor does not play an instrument. The conductor holds time and decides which voice speaks when. The card holds time. The card decides which nerve speaks now. The card is the conductor. The nerves are the players. The room is the symphony.
A second brain, in a shirt pocket, holding time for a body it was not born into.
SLIDE 6 — Exploded view of Gen 2: PCBA / SoM / VC / battery / chassis / belt clip
What is in here.
From the inside out — main board, NVIDIA Jetson Orin Super module in a SO-DIMM socket, vapor chamber and graphite spreader, two-cell lithium polymer battery, 6063-T5 anodized aluminum unibody chassis, belt clip on the back, USB-C and pogo magnetic ports on the side.
I’ll take each of those in turn.
Orin Super — the portable powerhouse
SLIDE 7 — Orin Super spec block + list of models running locally
The compute is an NVIDIA Jetson Orin Nano Super module. Sixty-seven TOPS, seven to twenty-five watts of configurable power envelope, sixty-nine by forty-five millimeters of footprint, paired with an integrated CUDA stack and TensorRT. I want to tell you what we run on it, because the spec sheet number doesn’t mean anything without the workload.
Concurrent, on the same module, in real time:
| - | YOLOv8 at thirty frames per second for VisionLink object detection |
| - | A CNN + LSTM stack on the eight-channel sEMG stream from Conductor — that’s the model that detects “are you snapping right now” |
| - | A CNN + Mamba sequence model for the L2 gesture classifiers — Mamba’s state-space backbone is what lets us hit sub-thirty-five millisecond inference on long-context wrist motion without the memory footprint of a transformer |
| - | And, when we want it, an eight-billion-parameter language model, locally, for natural-language device control |
4
Four models. One module. Zero cloud calls. We benchmarked this on the same hardware that is on the table behind me.
This is why we picked Orin. The previous generation of edge-AI wearables, including the products that did not survive last year, had two to six TOPS of compute on the wearer. They could not do this. The discontinuity is not incremental — it’s an order of magnitude.
SLIDE 8 — Stock dev kit vs. our custom carrier board, side-by-side
And we did not take Orin off the shelf.
The stock NVIDIA developer kit is one hundred millimeters by eighty millimeters of carrier board with a heatsink and a fan, designed for a desktop. We re-engineered the carrier — a custom four-layer PCB at half the footprint, the cooling moved into the chassis, the SO-DIMM exposed as a serviceable module, and the I/O wired out to three USB-C ports plus a magnetic pogo bus on the bottom edge for peripherals. What that means in practice: the wristband, the chest camera, the exoskeleton drive board, and the drone radio — every one of them plugs into a different port on this card without a dongle.
Connectivity — four languages to the world
SLIDE 9 — Connectivity diagram: Wi-Fi / BLE / UWB / USB-C / pogo
Four protocols, in order of how often we use them. Wi-Fi 6 for the drone and the smart lights. Bluetooth LE for the speaker and any low-power link to the wristband. Ultra-Wideband for three-dimensional wrist position — forty hertz, sub-ten-centimeter, the reason “point at the lamp” actually means a lamp. USB-C and pogo magnetic for the exoskeleton and the cameras — higher bandwidth, and the wire can carry power back into the card when we want it to.
Battery — long enough to matter
SLIDE 10 — Battery numbers in a single big stat block
Thirty-two-point-six watt-hours of two-cell lithium polymer, BMS-managed, USB-C PD at thirty watts. Three numbers: five hours continuous active load, eight hours realistic mixed load, sixteen hours standby with radios up and the L1 classifier listening. And one bonus number — the exoskeleton has a charging port that feeds the card back. Integrated wear time stretches past a normal workday.
Thermal — quiet and unfelt
SLIDE 11 — Thermal cross-section
You cannot wear something that sounds like a laptop. We abandoned active cooling early. What we settled on: a copper vapor chamber under the SoM, pyrolytic graphite spreading heat across the inside of the front face, and the chassis itself as the radiator. Twenty-five-watt peaks produce a surface rise of under twelve Kelvin against ambient. You can feel the card is warm. You cannot feel that it is hot. Mechanical engineering doing AI-model engineering’s job.
The agent on top — what makes this a platform
SLIDE 12 — AI Agent + 3-part open protocol
One more thing about the software stack, because it is the part that turns this from a product into a platform. Above the four inference models I just described, Orchestra runs a fifth thing. Not a perception model. An orchestration agent. Its job is not to recognize anything. Its job is to know what every connected nerve is producing, what every connected actuator is willing to consume, and to route between them in real time — semantically, not just electrically. This is the part of the system that lets us make an honest promise to a developer outside this company. If you have built a wearable, and you want it on Orchestra, we ask for three things. Three.
5
One — the interface. Channels, sample rate, units. What does the data look like coming off your sensor.
Two — the connection. BLE characteristic. Wi-Fi endpoint. USB CDC. Pogo bus. How does Orchestra physically reach you.
Three — the semantics. Heart rate in beats per minute. Hand position in centimeters. Oxygen saturation in percent. What does the data mean.
That is the whole protocol. Three contracts. No SDK to learn. No app to ship. No cloud to register against.
Three works because the data classes a wearable produces are not infinite. Physiological signals are time-series with a sampling rate and a unit. Kinematic signals are six-degree-of-freedom poses. Environmental signals are positions in space. The agent on the card already knows what those shapes look like — because Conductor and VisionLink already produce them.
Concrete example. A third party builds a blood-oxygen wristband. Interface — one channel of SpO2 at one hertz. Connection — BLE with a published characteristic. Semantics — saturation, seventy to one hundred percent. They publish those three. They ship.
The day after the SpO2 nerve attaches, on a job site, when a worker’s saturation drops below ninety-two percent, the exoskeleton slows down. Then it stops. Nobody wrote application code. The agent recognized a physiological signal class it already knew, and routed it to an actuator that was already listening.
You bring the wearable. We bring the brain that knows what to do with it.
SLIDE 13 — Conductor + VisionLink side-by-side
I want to be careful about how I describe what we built. Conductor and VisionLink are not “our two products” in the way that phrase usually lands in a launch presentation. They are reference implementations. They are the two nerves we built first, in-house, end-to-end, so we could prove the protocol on hardware we control before we asked anyone else to plug theirs in. Conductor is the reference nerve for muscle-and-motion. VisionLink is the reference nerve for vision-and-proximity. Both ship tonight. Both teach the next person who builds a nerve what the connector looks like. You have already seen both of them work — Conductor on this wrist, VisionLink on this chest. So I won’t re-explain what they do. Ninety seconds on each, on what’s under the surface.
Conductor — the nerve that reads you
SLIDE 14 — Conductor stack diagram
Eight surface EMG channels at one kilohertz. Six-axis IMU at two hundred. UWB ranging at forty. On a wristband that runs four hours between charges. The signal is the easy part. The model is the hard part. A two-layer classifier — a small always-on stage that detects activation and exit, and a per-device stage that loads on point-and-lock. Ten-second on-device adaptation — a short calibration, a fine-tune in place, no server in the loop.
Twitch-to-fired-command latency: thirty-three milliseconds. You watched this number tick on the dashboard while I was flying the drone. I am not arguing it from a slide.
VisionLink — the nerve that reads what you see
SLIDE 15 — VisionLink stack diagram
Chest-mounted camera at thirty frames per second. YOLOv8 on the Orin module. Per-frame classifier branch. Temporal smoother. Two modes — both in the reel. Active: finger count drives exoskeleton speed. Body to machine. Ambient: proximity and bearing of an approaching person drive graded haptic on the frame. Machine to body.
Orchestra is the first portable platform we know of that closes the body-machine loop on a single edge brain. Other camera wearables stop at a notification. Other exoskeletons stop at a control input. Orchestra does both directions, locally, on the person. If you know of another — find me after.
That is where we are tonight. Two nerves shipped. The platform under them ready for the next ones.
6
SLIDE 16 — 2×3 grid of future nerves, color-coded toC / toB
Now I want to look forward. Three time horizons, in the order we think about them. I’ll name them out loud so you can hold them in your head.
Horizon One — tonight, 2026.
One brain on the body. Two nerves attached. A handful of actuators in the room. The card is on my left rib. The system works. You have already seen it.
Horizon Two — two to three years.
Other people’s nerves attach.
The slide behind me has the five we are tracking most actively. I’ll go through them fast, because the pattern is what matters, not the list.
One. Eye tracking. Gaze + EMG + camera, stitched by the agent. The thing you look at is the thing your snap fires on.
Two. AR glasses — consumer. These. The prototype I have been wearing for the last twenty minutes. The dashboard you watched on the back wall has been rendering in front of my right eye the whole time. Same card. Same thirty-three milliseconds. The optics in this pair are not ours — we are the company that the optics plug into. The moment consumer AR glasses ship at scale, Conductor and VisionLink already render to them, because they were architected to. Today on a wall. Tomorrow on a retina. We’ve proved it works — it is on my face.
Three. Smart home — consumer. Conductor’s vocabulary, extended to thermostats, locks, blinds. One adapter per class.
Four. Robotics data — enterprise. Every wearer produces labeled áintent, motion, perception, outcomeñ. We license the pipeline.
Five. Drone operations — enterprise. One Tello on this table → coordinated fleet over a job site. The dispatcher does not care about count.
The point is not the five. None of those will be built by us. Every one attaches with the same three contracts — interface, connection, semantics. The agent does the rest.
Horizon Three — further out.
This is the part I am going to be most careful with, because the room has academics in it. The data Orchestra collects, while it is augmenting the wearer, is a continuous multi-modal record of you. Your kinematics. Your muscle activity. Your visual field. Your spatial position. The commands you intended. The outcomes you got. Time-aligned. Structured. Grounded in a known world model. Stored on the card, with the option of an encrypted twin in the cloud. That record, accumulating year over year, is a computational twin of the wearer. Let me be precise about what it is and isn’t. It is not consciousness. It is not personality. It is not anything science cannot define. It is a high-resolution, time-aligned approximation of what a particular human body does, sees, and wants — in a structured form that other intelligent systems can read.
Here is what the twin becomes useful for. When a humanoid in your kitchen needs to know what you mean by “put the plates away”, it does not need to call you. It queries your twin. When a fleet of drones over a worksite needs to verify whose authority a command was issued under, they check it against your twin.
When a remote actuator in another city wants to do work for you, the work is done against the interface your twin presents to it. You become the address. The work flows to it.
SLIDE 17 — VLA quadruplet diagram: vision + intent + action + state
A footnote for the academics, because the data shape I just described has a name in your literature. The field is converging on what’s called Vision-Language-Action models — VLAs — for embodied robotics. The bottleneck is not vision and it is not language. The bottleneck is action paired with the perception that caused it.
7
Look at what Orchestra produces every second a person wears it. Vision — the chest camera stream. Intent — the muscle signal, the closest thing we have to a user’s pre-motor command, ten milliseconds ahead of the actual movement. Action — the dispatched command, with timestamp and outcome. World state — the position of every actuator in the room, before and after.
Time-aligned. Structured. Grounded in a world model — because the dispatcher already knows what every command did, and the UWB layer already knows where everything was.
SLIDE 18 — World model: device map + person trajectory
Orchestra is, in the most literal sense, a world model with the human in it. The room is a graph. The actuators are nodes. The wearer is a moving anchor. Every gesture event is an edge. We do not have to label this data — the system labels itself as the wearer lives in it.
We are not racing to build the VLA. We are racing to build the data factory it has to learn from. The twin from Horizon Three is the same dataset, accumulated for one person, addressed by that person.
ACT 7 · Carbon and Silicon (1:45)
Three sentences, before the slide changes.
Today, the brain is on the body.
Tomorrow, the brain on the body and the brain in the cloud become continuous — same data, same agent, same protocol on both sides.
The day after that, your twin is the interface card the other intelligent systems use to find you.
SLIDE 19 — Carbon ⇄ Silicon transition
Twenty years from now there will be silicon-bodied creatures in your house. Not necessarily humanoid. Some will be. Some will be limbs. Some will be vehicles. Some will be drones over the field. All of them will be smart enough to act, and none of them — none of them — will understand a human being well enough to coexist gracefully without something in between. The thing in between is what we are building.
SLIDE 20 — Symphony metaphor
Carbon-based life — you, me, the wearer — emits intent through nerves we evolved over six hundred million years. Silicon-based machines emit action through nerves we have been building for eighty. The two nerve systems do not yet speak the same language. They will not, on their own.
A conductor is what turns independent voices into music.
SLIDE 21 — Orchestra wordmark + Move, the world follows.
We chose the name on purpose.
Tonight we have shown you the conductor and two of its players. The orchestra is what comes next — the players we have not built yet, and the music that gets made when carbon and silicon learn to keep time with each other.
Thank you.
8
Chief Executive Officer – Nan Zheng
01 . Opening
ON SCREEN — SLIDE 1 | MAY 28, 2026 — AUSTIN | You just saw it work. | Now let’s build it together.
Thank you, Bin. And thank you all for being here.
You just saw Orchestra working live — together with VisionLink and Conductor.
Today, I want to do two things.
Tell you who we want to build with.
And show you how builders can start working with us — today.
This is a product launch. It is also an invitation.
02 . Industries
ON SCREEN — SLIDE 2 | Industries we want to partner with.
So — who is this invitation for?
This opportunity is not limited to one category. These are the industries where Orchestra can create value.
If you are building in industrial automation, defense, smart home, or drones — we should talk. If you are working on AR/VR, automotive cockpits, service robotics, or humanoids — we should talk. If your company is in prosthetics, exoskeletons, rehabilitation, or Physical AI agents — we should definitely talk.
These are starting points, not boundaries. If your category is not on this slide, we are still welcomed to talk.
03 . Partnership
ON SCREEN — SLIDE 3 | PARTNERSHIP | What Orchestra adds to your product. | Efficiency · User Experience · Creativity
Now — if you’re in one of those industries, how would we actually work together?
For mature companies, Orchestra adds value in three ways.
Efficiency.
Faster workflows. Fewer mistakes. Less time managing screens. More time doing the work.
User experience.
Beyond buttons, menus, and touchscreens. Through gesture, movement, and intent. Your product understands your user better — and even responds before users have to ask.
Creativity.
The product ideas you couldn’t build before — because the interface wasn’t ready, the latency was too high, the devices couldn’t talk to each other, or the hardware was too clunky or too bulky — Now they can ship.
For partners, we’re not asking you to rebuild. We’re adding a layer — on top of what already works.
There are four ways to work with us.
Build on Orchestra — use our SDK and ship under your brand.
Build with us — joint development, shared roadmap.
Ship together — one product, both our names on it.
Create new categories — build something that one built yet.
Four ways to work with us. One principle: we’d rather be your partner than your vendor.
9
04 . Early Access Program
ON SCREEN — SLIDE 4 | ANNOUNCING TODAY | Orchestra Early Access. | Free. No equity. No exclusivity. No catch.
Partnership is one path. Early access is another.
Today, we are opening the Orchestra Early Access Program. It is built for startups, builders, small teams, and product creators working on what comes next. Selected teams will receive three things:
Hardware — our powerful Orchestra hub,
Software — Vision Link and Conductor pipeline, with the full SDK and engineering support.
Launch support — help you bringing the product story to market.
During early access, this is free. No equity. No exclusivity. No catch.
We want builders to explore what Orchestra can become in the real world.
The QR code on screen is the application page. Applications open today. We’re not selling Orchestra to you — we’re hoping to discover Orchestra with you.
05 . Closing
ON SCREEN — SLIDE 5 | In the mobile era, you looked down at the screen. | In this era, you look up at the world.
Let me close with this. In the mobile era, you looked down at the screen. In this era, you look up at the world.
That is the shift Orchestra is built for. From devices waiting for input — to systems that understand what people are trying to do.
If Orchestra fits what you are building — apply today. If you know someone we should talk to — make the introduction. The contact info is on the screen — info@wetourrobotics.com
Thank you for being here tonight — for witnessing this milestone with us.
Tonight, the door is open.
Let’s build what comes next — together.
Thank you.
10
FAQ
What did Wetour Robotics (WETO) announce in its Orchestra launch event?
Wetour Robotics introduced Orchestra as a wearable “portable external brain” that runs multiple AI models locally. It coordinates body-worn sensors, cameras, exoskeletons, drones and smart devices in real time, aiming to fuse human biological intelligence with machine intelligence in a single edge-computing platform.
What is the Wetour Robotics (WETO) Orchestra hub hardware and its key specs?
The Orchestra hub is a compact NVIDIA Jetson Orin Nano Super-based device delivering 67 TOPS of edge compute. It features a CNC aluminum unibody, carrier board optimized for wearables, Wi‑Fi 6, BLE, Ultra‑Wideband, USB‑C, and a battery designed for multi‑hour active and standby operation.
How does Wetour Robotics (WETO) use Conductor and VisionLink with Orchestra?
Conductor and VisionLink act as reference “nerves” for Orchestra. Conductor reads muscle and motion via sEMG, IMU and UWB on a wristband, while VisionLink uses a chest camera and detection models. Both feed signals into the Orchestra hub, which then drives exoskeletons, drones and other actuators.
What is the AI orchestration agent in Wetour Robotics (WETO) Orchestra?
The AI agent runs on top of Orchestra’s perception models and routes data between sensors and actuators in real time. Developers define interface, connection and semantics for their devices, and the agent uses this information to interpret signals and trigger appropriate actions without custom application code.
What is Wetour Robotics (WETO) Orchestra Early Access program?
The Orchestra Early Access program targets startups and builders working on new products. Selected teams receive Orchestra hardware, access to VisionLink and Conductor pipelines, a full SDK with engineering support, and launch assistance. The program is described as free, with no equity, exclusivity or hidden conditions.
How does Wetour Robotics (WETO) describe the future vision for Orchestra and Humanics?
Wetour describes a Humanics vision where human biological intelligence and machine intelligence operate as one system. Over time, data from Orchestra could form a computational twin of the wearer, helping robots, drones and other systems interpret human intent using structured, multimodal records of actions and context.