Related Experiment Video
Updated: Jan 11, 2026
08:25
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
9.5K
A Visual Dataset for Anomaly Detection in Self-Driving Laboratories
Shiwei Lin1, Zesen Chen2, Xiaobin Jia2
1Department of Computer Science and Technology, Tsinghua University, Beijing, China.
Scientific Data
|November 14, 2025
Summary
Researchers created a new dataset for anomaly detection in self-driving laboratories. This resource aids in developing intelligent systems for monitoring and decision-making in automated scientific experiments.
Area of Science:
- Robotics and Automation
- Artificial Intelligence
- Materials Science
Background:
- Self-driving laboratories (SDLs) enhance scientific discovery via automation, but anomaly detection remains difficult due to process complexity.
- Challenges include identifying anomalies in events, object states, and environmental conditions within SDLs.
Purpose of the Study:
- To introduce a novel dataset specifically designed for process anomaly detection in scientific experiments within an SDL context.
- To facilitate research in visual anomaly detection and automated monitoring systems for SDLs.
Main Methods:
- A dataset was constructed from a fully automated Polydimethylsiloxane (PDMS) synthesis workflow using collaborative robots.
- Images were captured from first-person perspectives using end-effector cameras on mobile and fixed robotic arms.
- The dataset comprises 1,671 images and 2,788 image-text pairs, including step-specific descriptions, anomaly labels, and region-level annotations across 11 checkpoints and 14 viewpoints.
Main Results:
- The dataset supports diverse visual anomaly detection tasks: image-text classification, anomaly type recognition, localization, grounded captioning, and attribution analysis.
- It provides a practical resource for developing intelligent monitoring and automated decision-making in SDLs.
Conclusions:
- This dataset is a valuable resource for advancing anomaly detection in automated scientific research.
- It enables the development of more robust and intelligent systems for self-driving laboratories, improving experimental reliability and discovery speed.
