Parameterized inverted and double pendulum model for controlling lower-limb active orthosis

Lower-limb active orthosis have been traditionally controlled by tracking clinical gait analysis (CGA) angle patterns. This approach however is very rigid and difficult to modify. This paper proposes a method based on the parameterization of simple dynamic models that explain human walking, which is more flexible and more intuitively modified than directly applied CGA patterns. In addition, a comparison between the angle trajectories obtained by this method and CGA is presented. In order to further test the method, it is applied to a simulated biped featuring the same mass and limb length distribution as a quadriplegic girl showing good results with the simplest possible controller, even in step-to-step transitions which are not explicitly considered by the parameterization. Finally, the joint coordinates used in the simulation were also implemented in the ATLAS prototype showing a natural looking gait.