Lateral balance control for robotic gait training.

For the rehabilitation of neurological patients robot-aided gait training is increasingly being used. Lack of balance training in these robotic gait trainers might contribute to the fact that they do not live up to the expectations. Therefore, in this study we developed and evaluated an algorithm to support lateral balance during walking, through controlling pelvis motions. This algorithm assists the pelvis, according to a natural pelvic sway pattern, rather than attracting it to the middle of the treadmill. The support algorithm was tested on six healthy young subjects who walked on a treadmill, while different support gains were introduced. Using a higher support gain resulted in a closer approximation of the pelvic sway towards the reference pattern. Step width and step width variability reduced when the external stabilization was provided, and the stability margin increased. This indicates that artificial stabilization reduces the need for active lateral balance control. The presented algorithm to support lateral balance provides a way to assist balance in a more physiological way, compared to attracting the subject to the centre of the treadmill. Here the user is attracted/assisted towards a more natural weight shift pattern. This also facilitates a more natural input of the load receptors, which are largely involved in the regulation of muscle activation patterns and the transitions between the different gait phases.