Abstract
We have developed a rehabilitation system for upper limbs, “PLEMO-P1”, which utilizes ER brakes as torque generators and is one of the passive haptic devices. To detect abnormalities of patients, we developed a new sensing device and installed it on the grip of the PLEMO-P3. In this paper, we establish effective training methods with PLEMO-P3. We compare several kinds of information from the sensor system (grasping force, reaction force against a plane, range of a wrist) and traditional assessment of stroke (BRS, FMA, SIAS and STEF). We could found effect of training and strong correlation between them. We suggest a training method with PLEMO-P3 depending on the results of the evaluation tests.