# Encord launches EMG-1, the biggest open dataset of muscle signals for robot training

> The new dataset pairs 300+ hours of forearm EMG recordings with first‑person video, cutting fingertip tracking error by 25%.

Oossa · 2026-10-20 · https://oossa.com/en/encord-launches-emg-1-the-biggest-open-dataset-of-muscle-signals-for-robot-train

Encord released EMG-1, an open‑source collection of more than 300 hours of synchronized muscle‑signal (EMG) and first‑person video data. The dataset is about 30 times larger than any similar public set. Tests showed that adding EMG to video lowered fingertip tracking error by 25 % and reduced errors during short visual gaps by 35 %. Researchers can download the data from encord.com/data‑labs.

## The facts

- 300+ hours of EMG and video recordings
- Released on 2026-10-20

## Why it matters

Robotics teams can improve hand‑pose tracking in occluded situations, leading to more reliable manipulation in industrial robots.

## Sources & references

1. [ENCORD RELEASES EMG-1, THE LARGEST OPEN DATASET LINKING MUSCLE SIGNALS TO ROBOTIC AI DEVELOPMENT](https://www.prnewswire.com/news-releases/encord-releases-emg-1-the-largest-open-dataset-linking-muscle-signals-to-robotic-ai-development-302914344.html) – PR Newswire RSS, 2026-10-20

Last updated: 2026-10-20