MIRA University of Twente
Department of Biomechanical Engineering

Lateral balance control during walking (Denise Engelhart)


Denise Engelhart


Z 122


Modeling and assessment of human balance control


Herman van der Kooij

Edwin van Asseldonk

Starting date:

January 2010

Denise Engelhart



Lateral Balance control during walking; prediction of foot placement


During rehabilitation after stroke, regaining walking ability is one of the major goals, as this greatly determines the level of socio-economic participation and the overall physical health of patients. There is growing scientific evidence that the largest functional improvement is due to task specific and intensive training of actively performed movements. Robotic devices can provide the patient with the necessary support during this training and are suitable for intensive and repetitive training. Besides this, they have a second major advantage with respect to regular therapy: they can be used for diagnostic measurements because every movement, force, torque and velocity can be measured. Using this data, therapists can diagnose the patient’s impairments and adapt the training program to the patient’s specific needs (assist-as-needed). Furthermore, the progress of therapy can be measured. This is useful for the patients’ recovery but also provides evidence for new training devices or programs.

In the past years an innovative robotic gait trainer (LOPES) was developed at the University of Twente. This robotic device provides support while the subject walks on a treadmill. One of its unique features is that the device is able to support movements in the frontal plane, for instance sideways movement of the pelvis and ad/abduction of the legs. This makes it possible to aid the subject in essential subtasks of walking like balance control and shifting the weight from one leg to the other during double stance.

Persons who have sustained a stroke usually present an abnormal and asymmetric gait pattern. This can be due to muscular weakness, spasticity, lack of coordination and decreased sensitivity. Another factor that highly influences the gait pattern of stroke patients is the disturbance in body balance. Especially frontal-plane balance is difficult during gait, because humans have a narrow base of support during the single stance phase and the medio-lateral displacement of the body produces imbalance. This imbalance must be actively controlled by foot placement, which is difficult for stroke patients.

LOPES training can help control foot placement, by using reference patterns of normal walking. This way it can assist in for instance step length and step height. For balance control, these reference patterns do not exists. Foot placement can differ between steps, due to obstacles and perturbations and if the robot has a fixed end point, the patient will fall. For training and assisting in balance control, the robot is not suitable yet. To make this possible, a model has to be developed which predicts foot placement during walking and includes reaction to perturbations.



Finding a model that predicts balance during walking in healthy patients.


Adapt this model so it becomes also valid for patients with impaired walking patterns, focusing on stroke patients.


Test in healthy subjects during treadmill walking if the model compares to the measurement outcomes. The experiment should include a way of unbalancing healthy subjects, so they compare to stroke patients.


If the model does not compare to the experimental outcomes, the model should be adjusted


First step is to develop a model that predicts foot placement. The model predictions are tested by performing experiments. Subjects walk on a treadmill and will be laterally perturbed by a linear motor. Foot placement will be detected by the VICON system. Afterwards the model predictions are compared to the experimental outcomes and the control laws will be adjusted so they can eventually be used in the LOPES to train balance in stroke patients.

Further information

No further information