Lower-limb rehabilitation robots often provide limited assessment of patient performance and personalized training. This paper proposes a conceptual framework that integrates multi-sensor data fusion with an adaptive Assistance-as-Needed (AAN) control approach. The framework consists of three steps: selecting clinically relevant parameters, fusing multi-sensor data into a performance score, and adjusting robotic assistance based on an adaptive AAN rule. This approach aims to improve therapy personalization, provide more informative feedback, and support evidence-based clinical decision-making. Future work will implement and evaluate the framework on a robotic gait trainer to assess its technical and clinical impact.
Shahrbaf et al. (Thu,) studied this question.