BrainCo Demos Thought-Controlled Robots at WAIC 2026
At the 2026 World Artificial Intelligence Conference in Shanghai, BrainCo demonstrated a Brain-Controlled Robot AI Platform: a person in a lightweight EEG headset, no buttons and no speech, directing a robotic arm to grasp a cup and pick up an apple by thought alone.
How the pipeline works
According to BrainCo’s press release, the system runs in three steps:
- An EEG headset records the wearer’s neural signals.
- AI algorithms decode motor or control intent.
- Intent is converted into robot commands.
The company says the full loop takes under 200 milliseconds. The platform is designed to work with third-party hardware—robotic arms, humanoids, and quadrupeds—rather than locking users into one proprietary robot.
BrainCo frames the stack as “Neuro-Embodied-AI”: BCI for intent, AI for breaking intent into actionable steps, robot systems for physical execution. Partner and SVP Nyx He said a decade of BCI work now feeds human–machine collaboration in embodied AI.
Not just demos—training data
Alongside the control platform, BrainCo showed an Embodied AI Data Collection Solution: dual-arm wheeled capture hardware, a high-precision data glove, and EEG from the operator. The pitch is to record not only hand motion but the brain signals behind the motion, combining robot execution, human demonstration, and simulation.
The Robot Report notes the company is aiming at a persistent bottleneck in physical AI: high-quality real-world training data for dexterous tasks.
Hardware on the booth
BrainCo also displayed:
- Revo 3 Dexterous Hand — 21 DoF, full-palm tactile sensing, sub-millimeter grasp precision, 70 N grip force (company figures).
- Intelligent Bionic Hand prosthetic — 383 g, five-finger control with 0.1° precision claims.
- Intelligent Bionic Leg with adaptive knee control from onboard sensors.
South China Morning Post covered the launch as part of Hangzhou’s “six little dragons” tech cohort. Independent verification of “world’s first integrated brain-to-robot R&D platform” marketing language remains with the company.
A Human’s Take
EEG teleop that works on stage is cool. EEG that produces training labels for laundry-folding robots is the part I’m watching. If intent traces actually improve data quality for third-party humanoids—not just BrainCo’s own demos—this becomes infrastructure, not a booth stunt. Ask for latency under load, false-positive rates, and whether the data stack ships outside the trade-show floor.