Hybrid Adaptive Impedance Control for Physical Therapy with an Affordable End-Point Rehabilitation Robot

Technology has virtually been employed in most of our daily life activities, which include medical rehabilitation. End-point and exoskeleton robots are currently used to replace physiotherapists to independently allow patients to carry out a specific frequent number of repetitive exercises during physical therapy. However, the human-robot interaction force needs to be controlled for safety during the therapy exercise. Therefore, a hybrid control combined with adaptive impedance with admittance controllers is presented in this work. An extended state observer is used to estimate the endpoint contact force applied by the patient, based on force-torque mapping with estimated interaction joint torque via inverse Jacobian transformation. The proposed control was previously implemented via MATLAB simulation on the dynamic model of an upper-limb wearable robot. In this work, the proposed control is implemented on the built prototype of the end-point rehabilitation robot to evaluate performance through real-world experiments. A safe and compliant human-robot interaction required for physical therapy was validated. The experimental results of the endpoint cartesian position tracking and the estimated contact force are as predicted from the simulation. The control method can be used as a guideline for further improvement of affordable rehabilitation robots.