Hilbert-Huang transform based pupil changes analysis for concentration assessment in skilled mowing

Area of Science:

• Human-computer interaction
• Biomedical engineering
• Ergonomics

Background:

• Mowing on steep terrain in Japan relies on manual labor.
• Complex visual environments impact worker concentration and pupil responses.
• Existing analysis methods may not fully capture nonlinear pupil dynamics.

Purpose of the Study:

• To propose a novel analysis method for human pupil changes during mowing on varied terrain.
• To identify specific pupil frequency patterns (IMFs) linked to mowing actions.
• To infer worker concentration status from pupil movement data.

Main Methods:

• Experiments conducted on flat and sloped terrain in Hiroshima, Japan.
• Application of Hilbert-Huang Transform (HHT) for action decomposition and nonlinear analysis of pupil data.
• Utilized a Multiple Comparisons and Filtering framework (MCFID) to identify relevant intrinsic mode functions (IMFs).

Main Results:

• Successfully identified IMFs directly related to specific mowing actions (cutting, lifting).
• Demonstrated that HHT is effective in analyzing nonlinear pupil changes.
• Established a method to infer concentration status by monitoring corresponding IMFs.

Conclusions:

• The proposed HHT-based method offers a more effective way to analyze pupil dynamics during complex tasks.
• Insights gained can improve training for lawn mower operators in challenging environments.
• The methodology provides a foundation for developing advanced fall detection systems.