CONTROL STRATEGY AND STRUCTURE IDENTIFICATION FOR EFFICIENT MOTION REHABILITATION

We propose control strategy for humans that have to perform goal-directed motion tasks and/or posture stabilization tasks. First, we define a set of variables that best characterize the dynamic performance of the controlled system in the required motion task. Second, driving forces/torques are properly assigned in order to achieve the required dynamic performance in an efficient way. The usual performance requirements are for positioning accuracy, movement response, and energy expenditure. For control purposes, we may use properly simplified dynamic models with a reduced number of degrees of freedom. It means that, with given motion task, we have to work with models that sufficiently well represent the kinematics, dynamics, and control functions. We can generate such models employing a general 3D computer model of the human body with 16 segments. Using simplified, yet realistic, models, we give two important examples to explain the basic features of our approach: upward posture stabilization and active ankle-foot orthosis.