A robotic system for delivering novel real-time, movement dependent perturbations.

Perturbations are often used to study movement control and balance, especially in the context of falling. Most often, discrete perturbations defined prior to the experiment are used to mimic external disturbances to balance. However, the largest proportion of falls is due to self-generated errors in weight shifting. Inspired by self-generated weight shifting errors, we created a novel, continuous mediolateral perturbation proportional to subjects' mediolateral center of mass movement with minimal delays. This perturbation was delivered by a robotic platform controlled by a real time Matlab Simulink model using kinematic data from a marker positioned at subjects' L5 as input. Fifteen healthy young adults stood as still as possible atop the robotic platform with their eyes closed. We evaluated the performance of the perturbation in terms of accuracy and delay relative to the input signal by using cross-correlations. The perturbations were accurate (r=-0.984), with delays of 154 ms. Such systematic perturbation significantly affected mediolateral sway, increasing its range (from 5.56±3.72 to 9.58 ±4.83 mm, p=0.01), SD (from 1.08±0.74 to 1.72±0.74 mm, p = 0.02), and mean power frequency (from 0.08±0.05 to 0.25±0.17 Hz, p<0.01). These perturbation characteristics enable inducing systematic, movement-dependent perturbations and open the door for future studies investigating self-generated movement errors.