Putting the human at the core of human-in-the-loop optimization: integrating clinical expertise and human capacities in exoskeleton learning. An interdisciplinary perspective paper.

Abstract

This interdisciplinary perspective aims to identify key limitations in current Human-in-the-Loop Optimization (HILO) approaches for lower-limb exoskeletons (LLEs) and to propose a shift toward more human-centred optimization. Drawing on literature from wearable robotics, rehabilitation science, motor learning, and clinical practice, we critically discuss three main limitations of current HILO approaches: narrow optimization scope, limited consideration of individual capabilities, and disconnection from personalized training. We argue that human-centered optimization should not only integrate physiotherapists' expertise but also expand their role as research partners in LLE development. Physiotherapists can contribute to adaptive control design, define user-centered outcome measures, and develop evidence-based training protocols that align human learning with technological adaptation. To support this shift, future research should prioritize: (1) using AI to interpret real-time, context-specific and user-specific data; (2) enabling physiotherapists to fine-tune device parameters based on individualized profiles; and (3) embedding structured, therapist-led training within HILO frameworks. Advancing human-centered HILO will require adaptable prototypes, multidimensional outcome measures, and sustained collaboration between engineering and rehabilitation science. This shift may help ensure that LLEs evolve into technologies that enhance mobility, autonomy, and participation in ways that are meaningful for users.