User preference-based human-in-the-loop tuning of exoskeleton assistance during walking

Area of Science:

• Biomedical Engineering
• Human-Machine Interaction
• Robotics

Background:

• Personalized exoskeleton assistance shows promise for improving human locomotion.
• Current methods for optimizing exoskeleton assistance are time-consuming and complex.
• User preference-based self-tuning offers a potential solution for individualizing assistance.

Purpose of the Study:

• To investigate a user-driven, human-in-the-loop tuning approach for personalized walking assistance using a hip exoskeleton.
• To determine if individuals can rapidly identify their preferred assistance settings through subjective perception.
• To assess the impact of personalized assistance on metabolic cost and sense of agency.

Main Methods:

• 11 healthy participants tuned four control parameters of a hip exoskeleton while walking on a treadmill.
• Participants self-selected their preferred assistance level.
• Metabolic cost and sense of agency were evaluated post-tuning.

Main Results:

• Participants found their preferred assistance settings in an average of 10.9 minutes.
• Preferred assistance profiles varied significantly between individuals.
• Metabolic cost was reduced by 16.6% with preferred settings, with robust benefits across timing variations.
• Sense of agency was sensitive to partial exoskeleton assistance.

Conclusions:

• User preference-based self-tuning is a viable and efficient method for personalizing exoskeleton assistance.
• This approach can significantly reduce metabolic cost during walking.
• Further research into guided tuning procedures may further optimize the individualization of exoskeleton assistance.