Home Robot Interaction Based on EEG Motor Imagery and Visual Perception Fusion

Area of Science:

• Robotics
• Artificial Intelligence
• Biomedical Engineering

Background:

• The global population is aging, increasing the need for assistive technologies for the elderly.
• Home robots offer potential for daily life assistance, but require advanced perception capabilities.

Purpose of the Study:

• To develop a multimodal human-robot interaction system for home robots to perceive elderly intentions and environments.
• To enhance the collaborative interaction between humans and robots in elder care settings.

Main Methods:

• Utilized Motor Imagery (MI) EEG signals with channel selection and Filter Bank co-Spatial Patterns (FBCSP) for classification.
• Integrated YOLO v8 for object detection and Machine Learning for scene recognition.
• Combined EEG classification with scene recognition to establish scene-intention correspondence for task recognition.

Main Results:

• Achieved a recognition accuracy of 83.4% for intention-driven task types.
• Demonstrated effective multimodal perception by integrating EEG and visual data.
• Validated the practical application value in human-robot collaborative interaction.

Conclusions:

• The proposed system provides a robust method for recognizing elderly intentions in home environments.
• This technology supports the development of smarter, personalized home assistance robots.
• The findings highlight the potential of multimodal perception in advanced human-robot interaction for elder care.