Systematic framework for performance evaluation of exoskeleton actuators.

Area of Science:

• Robotics
• Biomechanics
• Human-Robot Interaction

Background:

• Wearable devices, including exoskeletons, are increasingly utilized for mobility enhancement, rehabilitation, and industrial assistance.
• Optimizing actuator selection is critical for efficient exoskeleton design, traditionally relying on kinematic and dynamic analysis.
• Existing methods offer approximate actuator sizing but lack depth in analyzing performance, control, and cost implications.

Purpose of the Study:

• To present a novel learning-based evaluation method for analyzing exoskeleton actuation systems.
• To provide a systematic framework for assessing actuator performance and control algorithms.
• To improve the real-world operational efficiency of wearable robots through detailed analysis.

Main Methods:

• A real-world experimental setup was used to collect kinematic and dynamic data.
• Actuation system simulation focused on motor performance and control strategy development.
• Simulation results were experimentally validated, followed by real-world scenario testing.
• The framework replicates human-robot interaction kinematics and dynamics for comprehensive analysis.

Main Results:

• The developed learning-based method offers a detailed analysis of actuation systems.
• The systematic framework enables improved understanding of actuator performance and control strategies.
• Implementation on a back-support exoskeleton demonstrated substantial improvements in walking task performance.

Conclusions:

• The proposed learning-based evaluation method provides a robust framework for exoskeleton actuator analysis.
• This approach leads to enhanced operational efficiency and task-related performance in wearable robots.
• The study highlights the importance of detailed analysis for optimizing exoskeleton design and control.