ScholarWorks@중앙대학교

초록

Flat back syndrome is a condition characterized by the loss of lumbar lordosis due to weakened lumbarsupporting muscles, causing pain and inability maintaining the lumbar spine for extended periods. This condition leads patients to adopt incorrect postures in daily life, resulting in various inconveniences in activities of daily livings (ADLs). In this study, we aim to develop a soft wearable lumbar support robot that utilizes a tendon-driven system and the principle of three-point pressure to assist the lumbar function during daily activities. The developed robot includes a switch that enables the adjustment of the lumbar spine angle based on individual characteristics and specific situations. Performance evaluations demonstrated that the exosuit effectively assisted with trunk bending tasks, achieving an average range of motion from -2.13◦ to 37.87◦ in the unworn condition (without exo) and 0◦ to 23.75◦ when worn and activated (exo on), indicating improved flexibility. Electromyography (EMG) analysis revealed a significant reduction in activation of the thoracolumbar erector spinae (TES) and lumbar erector spinae (LES) during exo on conditions, with decreases of 9.95% and 10.36%, respectively, compared to the without exo condition. However, when the robot effectively assists lumbar extension and the spine is fully extended, lumbar flexion movements such as sitting induce tension in the wires, thereby restricting the user’s motion. Therefore, we developed a learning-based algorithm by attaching inertial measurement unit (IMU) sensors to both thighs to detect user intentions. The algorithm achieved validation and test accuracies of 94.12% and 92.31%, respectively, with high classification performance across most activities, though precision and recall were lower for dynamic transitional movements. This allows users to perform tasks involving extended and fixed lumbar positions more easily. We validated the algorithm by conducting verification assessments using test data and real-time evaluations of intention detection and motor control.

키워드