Performance Analysis of a Head and Eye Motion-Based Control Interface for Assistive Robots

Area of Science:

• Robotics
• Human-Computer Interaction
• Rehabilitation Engineering

Background:

• Assistive robots aid individuals with limited mobility, but many require residual arm/hand function.
• Hands-free control systems are crucial for individuals with severe mobility impairments, such as tetraplegia.
• Combining control modalities can enhance user experience and system performance.

Purpose of the Study:

• To develop and evaluate a novel multimodal control system for assistive robots using head and eye motions.
• To assess the performance of head motion (Magnetic Angular Rate Gravity - MARG sensor) and eye tracking for discrete and continuous control tasks.
• To investigate the usability of the system in a practical assistive scenario, such as supporting a drinking action.

Main Methods:

• A low-cost, compact multimodal sensor system combining a MARG sensor for head motion and an eye tracker for gaze detection was developed.
• Experimental evaluation included discrete button activation and 2D continuous cursor control (Fitts's Law task) with ten able-bodied subjects.
• A usability study was conducted with a collaborative robot assisting a drinking action, including one subject with tetraplegia.

Main Results:

• Able-bodied subjects showed no significant difference in activation time or throughput between head and eye motion control.
• Eye tracking resulted in a significantly higher error rate in the Fitts's Law task for able-bodied users.
• The subject with tetraplegia performed better with eye tracking for button activation and successfully used the system for the drinking task.

Conclusions:

• The developed head and eye motion control system is a viable hands-free solution for assistive robots.
• The system demonstrates potential for individuals with severe mobility impairments, including tetraplegia.
• Further research with a larger cohort of individuals with tetraplegia is recommended to validate these findings.