AgiBot, Chinese startup, pays workers to teleoperate robots for US data demand
Why does this matter now? Companies in the United States are scrambling for the kind of training data that lets autonomous systems move beyond simulation and act in the real world. The bottleneck isn’t the algorithms—it’s the hands‑on experience robots need to acquire.
That’s where a Chinese startup called AgiBot steps in. It runs a dedicated robotic learning center, hiring people to remotely control physical machines so AI models can pick up new capabilities. The model mirrors a growing trend: U.S.
firms are already outsourcing similar manual‑labeling work to workers in places like India, turning human effort into the raw material for machine learning. As demand for such data spikes, the economics of paying human operators to “teach” robots become a focal point for the industry. Jeff Schneider, a rob…
AgiBot has a robotic learning center where it pays people to teleoperate robots to help AI models learn new skills. Demand for this kind of robot training data is growing, with some US companies paying workers in places like India to do manual work that serves as training data. Jeff Schneider, a roboticist at Carnegie Mellon University who works on reinforcement learning, says that AgiBot is using cutting-edge techniques, and should be able to automate tasks with high reliability.
Schneider adds that other robotics companies are likely dabbling with using reinforcement learning for manufacturing tasks. AgiBot is something of a rising star within China, where interest in combining AI and robotics is soaring. The company is developing AI models for various kinds of robots, including humanoids that walk around and robot arms that stay rooted in one place.
AgiBot's AI-powered learning loop is precisely the kind of technology that US companies may need to master if they hope to reshore more manufacturing. A number of US startups are currently honing algorithms for new kinds of robo learning. These include Physical Intelligence, a heavily backed startup cofounded by some of the researchers who worked on the same project as Luo at UC Berkeley, and Skild, a spinout of Carnegie Mellon University that recently showed off robotic algorithms capable of adapting to new physical forms, including legged systems and robot arms.
China's huge manufacturing base is likely to give startups there some key advantages. These include a supply chain capable of prototyping rapidly and producing robots on a massive scale, a ready market for robot labor, and workers to help train robotic models. There are already more industrial robots operating in China than in every other country combined, according to the International Federation of Robotics, an industry body.
Will the approach scale beyond a single factory line? AgiBot’s method hinges on paying human operators to steer two‑armed robots while reinforcement‑learning algorithms absorb the motions, a combination that the company says accelerates skill acquisition. The pilot at Longcheer Technology, a manufacturer of smartphones and VR headsets, provides a concrete testbed, yet the report offers no data on error rates or throughput gains.
Meanwhile, a robotic learning center in Shanghai employs a workforce that teleoperates the machines, turning manual labor into training data. Demand for such data is rising, as noted by references to U.S. firms outsourcing similar work to India.
It's unclear whether the cost of paid teleoperation will offset the benefits of faster robot training, especially when compared to fully autonomous learning pipelines. The article stops short of detailing long‑term viability or the timeline for broader deployment. As the experiment proceeds, observers will likely watch for measurable improvements in manufacturing efficiency and for any signs that the model can be replicated in other production environments.
Further Reading
- AgiBot's New Industrial Humanoid Impresses with Archery Demo - Mike Kalil Blog
- China's Agibot takes on Tesla - The Rundown Robotics - The Rundown Robotics
- AgiBot Robotics Shined at IROS 2025 - the AgiBot World Challenge - OpenPR
Common Questions Answered
What is AgiBot's primary method for generating robot training data?
AgiBot operates a robotic learning center where it pays human operators to teleoperate two‑armed robots. The motions captured from these operators are fed into reinforcement‑learning algorithms, allowing AI models to acquire new skills more quickly.
Which company is providing the pilot testbed for AgiBot's robot training approach?
The pilot is being conducted at Longcheer Technology, a manufacturer known for smartphones and VR headsets. This partnership offers a real‑world factory line where AgiBot can evaluate the effectiveness of its teleoperated training data.
How does AgiBot's approach address the current bottleneck in autonomous system development?
The bottleneck is the lack of hands‑on experience for robots, not algorithmic limitations. By employing human operators to control physical machines, AgiBot supplies the real‑world interaction data that autonomous systems need to move beyond simulation.
What role does reinforcement learning play in AgiBot's robot training pipeline?
Reinforcement‑learning algorithms ingest the motions performed by teleoperated robots, using them to refine policies that automate tasks. According to Carnegie Mellon roboticist Jeff Schneider, this technique can achieve high reliability in task automation.
Why is there growing demand for robot training data from US companies, and how does AgiBot fit into this trend?
US firms are seeking large volumes of real‑world robot interaction data to train autonomous systems, often outsourcing manual data collection to workers in places like India. AgiBot meets this demand by providing a dedicated Chinese hub where paid operators generate the necessary training data.