Dynamic analysis and state estimation for wearable robotic limbs subject to human-induced disturbances

We present the Supernumerary Robotic Limbs (SRL), a wearable robot designed to assist human workers with additional arms and legs attached to the wearer's body. The SRL can work closely with the wearer by holding an object, positioning a workpiece, operating a powered tool, securing the human body, and more. Although the SRL has the potential to provide the wearer with greater strength, higher accuracy, flexibility, and dexterity, its control performance is hindered by unpredictable disturbances due to involuntary motions of the wearer, which include postural sway and physiological tremor. This paper presents 1) a Kalman filter approach to estimate the state of the SRL despite the involuntary wearer's motion, and 2) a method for improving the accuracy and stabilizing the human body and the SRL. The dynamics of the human-SRL system are analyzed, including human-induced disturbance models based on biomechanics literature. A discrete Kalman filter is constructed and its performance is evaluated in terms of error covariance. A "bracing" technique is then introduced to suppress the human-induced disturbances; one robotic limb grasps an environment structure and uses it as a support to attenuate the disturbances. We show how bracing can be used to shape the stiffness parameters at the robot base. This in turn allows to enhance state estimation accuracy in the areas of the workspace where the user needs assistance.