IO-AI Tech trains humanoid robots via teleoperation in Shenzhen
IO-AI Tech uses VR and motion-tracking to teleoperate 10 robot hands and gather training data for shelf-stocking and factory tasks.
TL;DR
- 01IO-AI Tech uses VR and motion-tracking to teleoperate 10 robot hands and gather training data for shelf-stocking and factory tasks.
- 02IO-AI Tech runs demonstrations and prototypes in Shenzhen, leveraging the city's manufacturing base to iterate hardware and integrations-system-integrations).
- 03The newsroom visit also included people using body-tracking sensors to control Unitree humanoids that mirrored operators inside a mocked-up apartment.
IO-AI Tech builds systems that let humans wear VR headsets, motion-tracking gloves and controllers to remotely operate humanoid robots for real-world tasks, and to collect teleoperation data that could train future autonomous agents. The startup, based about 45 minutes north of downtown Shenzhen, showed demos that included controlling 10 humanoid robotic hands at once and mapping a single operator's finger movements to all 50 robotic digits.
How does IO-AI Tech's teleoperation system work?
IO-AI Tech captures an operator's body and hand motions with VR headsets, handheld controllers and a custom motion-tracking glove, then maps those motions to different robot forms, while returning tactile feedback when available. The company demonstrated a glove that instantly transferred finger movements to 10 robotic hands, and the setup also allowed the operator to feel a ball placed in one of the electronic hands, showing two-way control and sensory feedback.
The startup's algorithms also perform shape and balance adjustments because a human and a robot "aren't always going to be the same shape, size, and weight," the company says, and some degree of autonomy is required to prevent robots from losing balance when they mirror human motions.
Where is this being tested and who is involved?
IO-AI Tech runs demonstrations and prototypes in Shenzhen, leveraging the city's manufacturing base to iterate hardware and integrations. The company invited visitors to control multiple robot hands and to try a VR-and-gripper system being tested by a Chinese convenience store chain for picking up boxes of medication and stacking shelves. The newsroom visit also included people using body-tracking sensors to control Unitree humanoids that mirrored operators inside a mocked-up apartment.
Local manufacturers are already collaborating with the startup. One partner named in the demos was Jack Sewing Machines, which is working with IO-AI Tech to train two-armed robots to perform tasks like ironing shirts. An executive from Jack Sewing Machines told the author these robots could fit onto an existing production line and automate work currently done by hand.
Why does IO-AI Tech collect teleoperation data?
Teleoperation serves two purposes for the company: it enables immediate remote operation in environments like factory floors and convenience stores, and it produces the focused training data AI researchers want for physical tasks. Si Chin, one of IO-AI Tech's cofounders, frames that data-gathering approach with a comparison: "It is similar to self-driving cars," she says, arguing that machines need task-specific training data to be deployed with increasing autonomy. The company also notes that robot teleoperation is gaining traction in some Chinese vocational schools.
What to watch
Watch whether teleoperation datasets from these Shenzhen pilots lead to robots taking on more autonomous roles on production lines, and whether Jack Sewing Machines moves from training to deploying two-armed ironing robots that can slot into existing lines. Another signal will be broader commercial tests by convenience store chains using the VR-and-gripper system for shelf-stocking and medication handling.
Written by The Brieftide · Source: Wired
The Brieftide Daily · 06:00
Briefs like this one, in your inbox every morning.
Continue reading
More in AI InfrastructureIEEE launches virtual training course on large language models
IEEE is offering a virtual training course that teaches engineers to use large language models as reasoning engines in development.
AI4SE and SE4AI: A decade review of AI in systems engineering
H. Sinan Bank, Daniel R. Herber and Thomas Bradley map three research phases and assess 1.
Amazon's AWS may sell Trainium chips to challenge Nvidia
AWS executives say selling Trainium to third parties is possible, with Andy Jassy estimating a potential ~$50 billion annual run rate.
Hyperscalers AI spending to outpace cash flow by Q3 2026
Epoch AI data shows infrastructure spending growing ~70% annually versus operating cash flow at ~23%, with a crossover around Q3 2026.