Towards postural balance control of exoskeletons

Abstract

Lower-limb wearable exoskeletons have been designed to assist people that have a spinal cord injury during standing and walking. However, because these people generally also have impaired balance, it is difficult, if not impossible for them to operate these exoskeletons without additional supporting aids, such as crutches. Ideally the exoskeleton supports its user’s balance, preferably in a human-like way to match the user's natural intention. Therefore, proper balance control of the exoskeleton is required. This work presents the first steps taken towards postural balance control of lower-limb wearable exoskeletons. The focus is specifically on standing balance control strategies for exoskeletons, inspired by human and humanoid standing balance. The first goal of this thesis was to explore balance control strategies for the application in a lower-limb exoskeletons, with a particular focus on human-like motion generation. In Chapter 2 the ability of the momentum-based controller to generate human-like feet-in-place balance recovery strategies was investigated. Besides feet-in-place balance recovery strategies, people also use a reactive stepping strategy to maintain balance. Therefore, in Chapter 3 it was investigated whether the occurrence of reactive stepping could be predicted using a classification-based method, and what features are most relevant for that prediction. The second goal of this thesis was to verify the effectiveness of exoskeleton balance support. Hence, the effects of an ankle exoskeleton and an ankle-knee exoskeleton on the balance of able-bodied users and (three) users with an incomplete spinal cord injury respectively were assessed in Chapters 4 and 5. By modeling human balance for the use in an exoskeleton on the one hand, and by analyzing and implementing existing balance control strategies on the other, the results presented in this thesis provide insight into how to impose standing balance on exoskeletons and their users.