Human Torque Estimation for an LMI-Based Convex Control Rehabilitation Strategy Using Assistive Robots

The number of people affected by motor impairment has increased considerably in recent years, raising, in consequence, the interest in the use of assistive robotic devices for additional motricity and rehabilitation purposes. The main challenge for this kind of devices is dealing with the interaction between the human and the robot for a cooperative movement, especially in active rehabilitation schemes, where patients are incited to participate in the movement tasks, in order to increase efficiency of the rehabilitation protocol. A question that arises in this specific context is how to estimate human contribution in order to personalize the rehabilitation tasks. Current solutions being based on cumbersome measurement devices, this paper suggest an observer-based solution, allowing to estimate the human torque based on few dynamical measurements, in addition to an LMI-based computed-torque controller. This suggested scheme has been tested and validated using OpenSim software, which is widely used for biomechanical modeling, simulation and analysis.