Chien, Li ChienLi ChienChienChen, DianfuDianfuChenCHAO-CHIEH LAN2025-09-242025-09-242016https://www.scopus.com/inward/record.uri?eid=2-s2.0-84992344975&doi=10.1109%2FAIM.2016.7576780&partnerID=40&md5=608350102744e623419e079959556b94https://scholars.lib.ntu.edu.tw/handle/123456789/7324452016 IEEE International Conference on Advanced Intelligent Mechatronics, AIM 2016. Banff; AB. conference code:124072After-stroke rehabilitation of patients with shoulder disabilities traditionally requires repeated and progressive training exercises. Complete recovery is usually costly and lengthy. Aimed at reducing the cost and time of shoulder rehabilitation, powered exoskeletons are developed to provide automatic and steady rehabilitation training. Our previously developed shoulder exoskeleton is small-size, light-weight, and requires low power consumption. This paper further improves its wearability and safety by introducing an adaptive mechanism. This adaptive mechanism can compensate for the misalignment between exoskeleton and human upper limb and size variation among different subjects. To ensure safe interaction, a compact two-axis force sensor will be realized in the adaptive mechanism to obtain accurate force at the human-exoskeleton interface. Combined with the series elastic input motors, the exoskeleton can provide accurate force control and avoid human injury. Bidirectional actuation between exoskeleton and human can also be allowed, which is required for various rehabilitation processes. We expect this shoulder exoskeleton can provide a means of automatic shoulder rehabilitation.Adaptive MechanismAxis MisalignmentGravity Balancing MechanismMulti-axis Force SensorShoulder RehabilitationSpherical MechanismUpper Limb ExoskeletonAlignmentIntelligent MechatronicsNeuromuscular RehabilitationPatient RehabilitationAdaptive MechanismAxis MisalignmentsGravity-balancingMulti-axisShoulder RehabilitationsSpherical MechanismsUpper LimbsExoskeleton (robotics)Design of an adaptive exoskeleton for safe robotic shoulder rehabilitationconference paper10.1109/AIM.2016.75767802-s2.0-84992344975