LAUSR

Gait analysis is commonly addressed through inverse dynamics. However, forward dynamics can be advantageous when descending to muscular level, as it allows activation and contraction equations to be integrated with motion thus providing better dynamic consistency, or when studying assisted gait, as it enables the estimation of the interaction forces between subject and devices even if the motion capture process doesn’t provide enough resolution to distinguish the motions of limb and device. Control-based methods seem to be the most natural choice to carry out the forward-dynamics analysis of an acquired gait, but several options exist in their application. The paper explores such options for healthy and assisted gait, and concludes that the computed torque control of all the subject’s degrees of freedom is the alternative that provides the most accurate results. Moreover, the study of its more problematic underactuated variant accompanied by contact models showed to be connected to neighbor challenging topics as gait prediction or walking simulation of humanoids.