Optimization of Real-Time EEG Artifact Removal and Emotion Estimation for Human-Robot Interaction Applications
Mikel Val-Calvo1,2, José R Álvarez-Sánchez2, Jose M Ferrández-Vicente1
1Departamento Electrónica, Tecnología de Computadoras y Proyectos, Universidad Politécnica de Cartagena, Cartagena, Spain.
Frontiers in Computational Neuroscience
|December 19, 2019
Summary
This study optimizes real-time emotion estimation using electro-oculography (EOG) by improving artifact removal and feature extraction. The new method achieves high accuracy for affective computing, enabling better human-robot interaction.
Area of Science:
- Affective Computing
- Human-Robot Interaction
- Biomedical Signal Processing
Background:
- Real-time emotion estimation is crucial for affective human-robot interaction.
- Current methods face challenges in optimizing accuracy and processing time, especially with artifact removal.
Purpose of the Study:
- To propose an optimized methodology for real-time emotion estimation.
- To address challenges in artifact removal, feature extraction, and classification for electro-oculographic (EOG) signals.
- To validate the methodology's performance under real-time constraints and varying experimental paradigms.
Main Methods:
- Compared two real-time electro-oculographic (EOG) artifact removal techniques based on information loss and processing time.
- Developed an emotion estimation methodology using stable features, selected electrodes, and smoothed feature spaces.
- Tested the methodology on the SEED database for discrete emotional states (three affective states).
Main Results:
- The proposed methodology effectively removes artifacts while preserving information and reducing processing time.
- Achieved high accuracy in real-time emotion estimation under both subject-dependent and subject-independent conditions.
- Demonstrated the method's suitability for real-time affective computing applications.
Conclusions:
- The optimized emotion estimation methodology meets real-time constraints with high accuracy.
- This work advances the field of affective human-robot interaction through efficient EOG signal processing.
- The proposed approach provides a robust solution for real-time emotion recognition.
