Published November 10, 2026 | Version v2

Dataset of Workers Physiological Signals During Assembly Tasks

  • 1. ROR icon Tampere University

Contributors

  • 1. ROR icon Tampere University

Description

Description

The physiological signals in this study were collected from four operators working on a real-world assembly line. The line consisted of one manual workstation and one collaborative workstation involving human-robot interaction. The setup was provided by Silverline Appliances Co., an industrial partner in the AI-PRISM project. Silverline manufactures home appliances, and the experiment took place at the labeling and testing stations of the hood assembly line. The first workstation focused on functional tests to verify product operability, while the second handled cleaning and labeling of the hood.

The dataset includes signals derived from cardiovascular and ocular activity. Cardiovascular signals were recorded using an ECG chest band, and ocular signals were captured via eye-tracking glasses.

Data files

Each .zip file corresponds to one of the four operators. Inside each .zip, you will find four folders: blinks, fixations, saccades, and ECG data. These folders contain .csv files from multiple recording sessions. Each .csv file is organized into columns with features specific to its category (blinks, fixations, saccades, ECG) and generic identifiers such as:

  • Worker ID
  • Experiment Name
  • Event Label
  • Event Group
  • Start
  • End
  • Timestamp

The Event Label column indicates the task the operator was performing at a given moment. For eye-tracking data, it may also include information about the stimulus the operator was looking at (e.g., task_name or worker(colleague)).

Similarly, the Event Group column primarily identifies the product numbers handled during that recording session. For eye-tracking data, it can also indicate whether the operator was looking at a product or another worker.

For ECG data, the Start and End columns mark the beginning and end of the event described in Event Label, while Timestamp specifies when the data was collected.

Additionally, each NASA_TLX_Calculator_00X.xlsx file contains the NASA-TLX data corresponding to each recording in which each of the operators took part.

Files

001.zip

Files (12.7 MB)

Name Size
md5:fc1bbd572bf0296f4ab64dd5c7b07efa
3.4 MB Preview Download
md5:a4e1b03d6648eae7a65c65c38689ffde
3.9 MB Preview Download
md5:cd8bb2c5da8364e69796a3937c0f4f81
3.3 MB Preview Download
md5:6235370ac7f2dcee2e9687f81ac5f776
1.6 MB Preview Download
md5:a8a8394d9e4a78d2322b150634834be8
137.7 kB Download
md5:7e48319f5945786501cac2bc2a4b6d37
137.6 kB Download
md5:b9dca93e253d961b4f9bc157bf0e734a
137.0 kB Download
md5:adeaceb25cb7e2330cf96b9d851e66ff
137.0 kB Download

Additional details

Related works

Is described by
Journal: 10.12688/openreseurope.24614.1 (DOI)
Is supplemented by
Dataset: https://zenodo.org/records/15340717 (URL)

Funding

European Commission
AI-PRISM - AI Powered human-centred Robot Interactions for Smart Manufacturing 101058589

Dates

Available
2025-11-20