Humanoid robots may not yet be capable of independently handling household chores or warehouse operations, but a startup in Shenzhen, China, is demonstrating a practical way to bridge that gap. IO-AI Tech has created a teleoperation platform that allows human workers to remotely control humanoid robots performing real-world tasks across the region. While the robots appear autonomous to observers, human operators remain behind the scenes, using virtual reality and motion-tracking technologies to guide their actions, tells Wired.com (full article available to subscribers).
At the company’s offices, employees wear VR headsets, handheld controllers, and motion-tracking equipment to take command of robots built by multiple manufacturers. The platform is designed to be hardware agnostic, enabling operators to seamlessly switch between different robot models. Although the robots rely on humans for complex manipulation, they retain enough onboard intelligence to maintain balance and execute basic movements, reducing the operator’s workload.
The remotely controlled robots are already performing a variety of practical jobs. In residential settings, they wash dishes, fold laundry, and complete other domestic chores. In commercial environments, they restock convenience stores, move goods, and operate equipment inside warehouses. This approach allows businesses to deploy humanoid robots today without waiting for artificial intelligence to achieve complete autonomy.
However, the teleoperation system serves a larger purpose than simply enabling remote labor. Every action performed by a human operator generates valuable real-world training data. By recording these interactions, IO-AI aims to build increasingly capable AI models that will eventually allow robots to complete the same tasks independently. In this sense, today’s human operators function as both workers and trainers, providing the demonstrations that future autonomous systems will learn from.
The company’s strategy highlights a growing trend in robotics development. Instead of waiting for fully autonomous humanoids to become reliable enough for everyday use, developers are combining human expertise with robotic hardware to accelerate learning. This hybrid approach could significantly shorten the path toward capable embodied AI while creating a new category of technology-enabled blue-collar work.
For now, humans remain an essential part of the control loop. As AI models mature through continuous data collection, teleoperation may gradually shift from a permanent solution to a stepping stone toward fully autonomous humanoid robots.