Related Experiment Video
Updated: May 16, 2025
09:41
Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
Published on: April 21, 2023
1.4K
Computer vision and tactile glove: A multimodal model in lifting task risk assessment
Haozhi Chen1, Peiran Liu1, Guoyang Zhou1
1Purdue University, West Lafayette, IN, USA.
Applied Ergonomics
|April 2, 2025
Summary
This study combines tactile gloves and computer vision to assess lifting injury risks, offering a new method for occupational safety. The developed model accurately predicts risk levels, aiding in preventing workplace musculoskeletal injuries.
Area of Science:
- Occupational Safety and Ergonomics
- Biomechanics
- Human-Computer Interaction
Background:
- Work-related injuries from overexertion, especially lifting, pose significant occupational safety challenges.
- Traditional assessment tools like the Revised NIOSH Lifting Equation (RNLE) demand extensive training and practice.
- Existing single-modality assessment methods have limitations in real-time, comprehensive risk evaluation.
Purpose of the Study:
- To develop and validate a novel approach integrating tactile gloves and computer vision (CV) for enhanced assessment of lifting-related injury risks.
- To address the limitations of traditional methods by creating a more accessible and efficient risk assessment tool.
- To investigate the efficacy of combining hand pressure and 3D body pose data for predicting lifting risk categories.
Main Methods:
- Thirty-one participants completed 2747 lifting tasks across defined risk categories (LI < 1, 1 ≤ LI ≤ 2, LI > 2).
- Data collected included hand pressure via tactile gloves and 3D body poses estimated using CV algorithms from video recordings.
- A Convolutional Neural Network (CNN) model was developed using the combined feature data for risk prediction.
Main Results:
- The CNN model demonstrated a high overall accuracy of 89% in classifying the three lifting risk categories.
- The integration of tactile glove data and CV-derived body pose information proved effective for risk assessment.
- The study successfully established a predictive model for lifting-related injury risks.
Conclusions:
- The developed system shows significant potential as a real-time, non-intrusive tool for assessing lifting-related injury risks.
- This approach can assist ergonomic practitioners in proactively mitigating musculoskeletal injury risks in various workplace environments.
- The fusion of tactile sensing and computer vision offers a promising advancement in occupational biomechanics and safety.
