中 / EN
2026-07-26 00:00 Industry News Source language EN

Zander Labs and Encord Trial EEG Data for Robot Model Training

Summary Zander Labs and Encord are piloting the collection of first-person video alongside workers’ EEG to build robot-training data. Encord plans to test an initial dataset on customer models before deciding whether to scale collection.
Why it matters The decisive test is whether adding EEG improves model performance beyond what first-person video alone provides.

Zander Labs与Encord试用脑电数据训练机器人模型
Image: Zander Labs

BCIwiki (bciwiki.com) — Zander Labs and Encord are piloting the use of EEG in robot-training data, according to partnership information and TechCrunch reporting. Encord plans to build an initial brain-wave-labelled dataset, test it on customer models and use those results to decide whether to scale collection.

TechCrunch described worker Andrew Ceja playing Jenga at Encord’s facility in San Leandro, California while wearing a device with a camera and EEG sensors. The camera records a first-person view while sensors measure brain activity. German company Zander Labs made the device.

Zander Labs hopes to infer mental states associated with errors, intentions and surprise, adding information that may be difficult to obtain from video alone. Company neuroscientist Lukas Gehrke said activity at different task stages could also help developers decide when to invoke more computationally demanding models.

Encord provides data labelling and model evaluation tools and also collects first-person video and robot teleoperation data. This pilot concerns training-data construction and evaluation; it has not yet established that adding EEG improves robot capabilities.

Compiled by BCIwiki from public sources

Sources · 2
techcrunch.com 2026-07-26
zanderlabs.com 2026-07-26
Read original ↗
Suggest a correction Revisions · none 中 / EN
© 2026 BCIwiki.com Digest Topics Tips Subscribe Revisions About