Intentional Target Reaching With an Underactuated Exoskeleton in a Cluttered Environment

Area of Science:

• Robotics
• Human-Computer Interaction
• Biomechanics

Background:

• Upper-limb exoskeletons offer potential for individuals with arm weakness.
• Accurate user intent detection is crucial for effective exoskeleton assistance.
• Current systems struggle with precise intent recognition in complex environments.

Purpose of the Study:

• To develop and validate an Intentional Target (IT) estimation framework for precise user intent recognition.
• To enable effective control of portable, underactuated upper-limb exoskeletons for daily activities.
• To improve task success rates in real-world scenarios involving cluttered environments.

Main Methods:

• Integrated shear force-based forearm rotation detection, vision-based object recognition, and hand tracking for IT estimation.
• Utilized a fusion model combining Human-Robot-Objects (HRO) model with human behavior data for joint angle estimation.
• Employed estimated joint angles as benchmarks for controlling portable underactuated exoskeletons.

Main Results:

• The IT estimation framework demonstrated precise target identification in cluttered tabletop environments.
• The system achieved a 92% task success rate (TSR) in trials with healthy adult participants.
• 100% accuracy was observed for specific object interactions, such as reaching for a cup.

Conclusions:

• The proposed IT estimation framework is effective for identifying user intent in complex, real-world settings.
• The control strategy successfully addresses underactuation limitations in portable exoskeletons.
• This approach enhances the reliability and effectiveness of reaching assistance for individuals with arm weakness.