Combined Estimation of Friction and Patient Activity in Rehabilitation Robotics

Robot-assisted gait training has shown promising results in inpatient rehabilitation. However, the progress the patients can achieve during this period often stagnates or even deteriorates after their discharge. To extend the robot-assisted gait training to the patient’s home area, a new rehabilitation robot is developed. Since the home training is not supervised by physiotherapists, a challenging training is needed as well as a reliable and comprehensible feedback to the patient. Therefore, the patient activity has to be estimated and set apart from the occurring friction forces falsifying the estimation. In this paper, an approach is described that offers a combined estimation of friction forces and patient activity, based on a new dynamic friction model and a central difference Kalman filter.