Encord and Zander Labs Pilot Brain Wave Data to Train Physical AI and Humanoid Robots, Addressing Data Bottleneck

Encord is testing brain wave sensors from Zander Labs as a new data modality to train physical AI and humanoid robots, aiming to address the industry's significant data bottleneck by capturing human mental states directly.

Capturing Human Intent with Brain Waves

The core of this initiative involves human 'pilots' wearing electroencephalogram (EEG)-equipped headsets while performing various manipulation tasks. As these tasks are carried out, the headsets record brain activity, allowing researchers to deduce critical mental states such as error detection, intent, and surprise. This rich, real-time data provides a novel signal that could significantly improve how AI models understand and react to complex physical environments.

Encord plans to develop an initial dataset tagged with these brain wave signals. This dataset will then be rigorously tested against existing customer robotics models to evaluate whether this new data modality meaningfully improves performance. If successful, this method could offer a valuable signal for deploying high-effort models, pushing the boundaries of what physical AI can achieve.

Addressing the Physical AI Data Bottleneck

The physical AI industry, particularly in the development of humanoid and warehouse robots, faces a fundamental challenge: a severe data bottleneck. Training these advanced systems requires vast amounts of expensive, purpose-manufactured real-world data. Unlike digital AI, which can use massive online datasets, physical AI demands data that reflects intricate real-world interactions and nuances.

Vineeth Velmurugan, Encord's head of robot learning and a veteran of OpenAI's robotics lab and Berkshire Grey, estimates that a dataset roughly five times the size of YouTube's entire video corpus will be necessary to truly break through this data barrier. This highlights the immense scale of data required and the urgency of finding more efficient and effective data generation methods.

The Role of Data Generation as a Business

This trial with Zander Labs underscores a broader trend: data generation itself is evolving into a standalone business, moving beyond being merely a research problem. Companies like Encord are not just building AI models; they are also developing sophisticated infrastructure and methodologies for collecting and annotating the specific types of data needed to train these models. Encord already operates extensive egocentric video data collection and teleoperated robot data facilities across multiple global locations, further solidifying its position in this critical area.

Implications for General-Purpose Robots

If brain wave tagging proves effective, its implications for the future of robotics are substantial. By providing AI models with a deeper understanding of human mental states during task execution, it could accelerate the timeline for creating truly capable general-purpose robots. These robots would be better equipped to handle unforeseen situations, learn from mistakes more efficiently, and adapt to new tasks with greater autonomy, ultimately leading to more robust and versatile AI systems.

What to Watch Next

The collaboration between Encord and Zander Labs represents a significant step in exploring new data modalities for AI training. The success of this initial brain wave-tagged dataset will be a key indicator of its potential to revolutionize physical AI development. As the industry continues to grapple with the challenges of real-world data collection, innovative approaches like this will be crucial for advancing the capabilities of humanoid and warehouse robots. Readers interested in the latest AI updates can find more information on AI news and top AI tools.

Sources

• A Review of Brain-Computer Interface Technologies: Signal Acquisition Methods and Interaction Paradigms
• Large-Scale AI and Foundation Models for Neuroscience
• https://techcrunch.com/2026/07/26/are-brain-waves-the-next-unlock-for-physical-ai/
• Brain.space remakes the EEG for our modern world (and soon, off-world) | TechCrunch
• Instagram