Predicting targets of human reaching motions using different sensing technologies.

Area of Science:

• Human-Computer Interaction
• Biomedical Engineering
• Machine Learning

Background:

• Predicting voluntary motions is key for intuitive human-computer interaction (HCI).
• Limited comparative studies exist on sensing technologies for motion prediction.
• Evaluating different modalities is crucial for HCI system design.

Purpose of the Study:

• To compare the predictive performance of various sensing technologies for human reaching motions.
• To investigate the efficacy of combining different sensing modalities.
• To provide recommendations for selecting sensing technologies in HCI applications.

Main Methods:

• Utilized supervised machine learning for motion prediction using electroencephalography (EEG), electrooculography, eye tracking, electromyography (EMG), and hand position.
• Assessed prediction accuracy at different time points (pre- and during-motion).
• Developed an algorithm to combine modalities based on information availability over time.

Main Results:

• Electroencephalography (EEG) predicted motion pre-onset but required subject-specific training and decreased with more targets.
• Electromyography (EMG) and hand position offered high accuracy post-motion onset.
• Eye tracking demonstrated robust, high accuracy at motion onset.
• Combining modalities, including contextual data, showed significant advantages.

Conclusions:

• No single modality is optimal for all human-computer interaction scenarios.
• Eye tracking and combined modalities show promise for early and accurate motion prediction.
• Recommendations are provided for selecting sensing technologies based on application needs and performance criteria.