Related Experiment Videos
Design of an Adaptive Human-Machine System Based on Dynamical Pattern Recognition of Cognitive Task-Load
Jianhua Zhang1, Zhong Yin2, Rubin Wang3
1Intelligent Systems Group, School of Information Science and Engineering, East China University of Science and Technology Shanghai, China.
Frontiers in Neuroscience
|April 4, 2017
Summary
This study developed an algorithm to classify operator cognitive task-load (CTL) using EEG and ECG data. The system adaptively allocates tasks to maintain optimal operator performance in human-machine systems.
Area of Science:
- Human-Computer Interaction
- Cognitive Engineering
- Biomedical Signal Processing
Background:
- Safety-critical systems require monitoring operator cognitive task-load (CTL).
- Maintaining optimal CTL is crucial for sustained performance and safety.
- Existing methods lack dynamic, real-time CTL assessment.
Purpose of the Study:
- To develop a cognitive task-load (CTL) classification algorithm for adaptive human-machine systems.
- To implement a strategy for sustaining optimal operator CTL levels over time.
- To improve overall human-machine system performance through adaptive automation.
Main Methods:
- Developed a non-linear dynamic CTL classifier using electroencephalogram (EEG) and electrocardiogram (ECG) features.
- Employed Least-Squares Support Vector Machine (LSSVM) for dynamic pattern classification.
- Reduced 56 initial features to 12 salient features using Locality Preserving Projection (LPP).
Main Results:
- Achieved an 80% correct classification rate for a 5-class CTL problem.
- Constructed participant-specific dynamic LSSVM models for instantaneous CTL classification.
- Demonstrated improved human-machine system performance via adaptive task allocation.
Conclusions:
- The developed CTL classification algorithm and adaptive strategy effectively sustain optimal operator cognitive load.
- Real-time CTL monitoring and adaptive automation enhance human-machine system performance and safety.
- The LSSVM-based approach with reduced feature sets offers a viable solution for dynamic CTL assessment.