Back to the Future: Towards versatile neuromechanical control of back-exoskeleton support

Low-back pain (LBP) is a highly prevalent and workers commonly exposed to high low-back loads are at greater risk of developing LBP. Back-support exoskeletons can be a versatile tool to reduce low-back loading, as they can take over part of the lumbar moment generated by the trunk extensor muscles. However, current exoskeletons relying on actuators (active exoskeletons) may limit the user’s intended movement, hampering acceptance, and current exoskeleton control may not leverage actuator versatility. This thesis aims to improve active exoskeleton design to allow intended movement, implement intelligent control accounting for the distribution between actively and passively generated lumbar moments, and explore using trunk extensor muscle fatigue development estimations for exoskeleton control. In chapter 6, we showed that an active back-support exoskeleton design including two sets of bilateral actuated joints (mid-lumbar and hip joint) allows intended movement, unlike designs including only one set of bilateral actuated hip joints. Furthermore, in contrast to existing exoskeleton control approaches, we showed that subject-specific exoskeleton control based on the actively generated lumbar extension moment due to gravitational forces acting on the upper body yielded a decreased support at higher lumbar flexion angles, preventing counter-productive support when active moments decrease due to increasing passive lumbar extension moments. Throughout chapters 2, 3, 4 and 7, we found that trunk extensor muscle fatigue development can be accurately estimated during prolonged bending using high-density electromyography (HDsEMG). Average motor unit action potential conduction velocity (CV) from HDsEMG is useful if enough CV estimates are obtained (chapter 2 & 3). As a more practical alternative, HDsEMG spectral content yielded high correlations with muscle endurance if a subset of channels was selected based on the magnitude of electromyographic manifestations of muscle fatigue (chapter 7). Conventional surface electromyography, however, may not be suitable due to localized electromyographic manifestations of muscle fatigue (chapter 3 & 4). Early prediction of the limit of endurance when using HDsEMG requires further research (chapter 7). In chapter 4, we showed that intermittent changes in trunk extensor muscle length can reduce the rate of muscle fatigue development compared to maintaining constant muscle length and self-selecting muscle length. Such intervention could be implemented as a control strategy in an active exoskeleton including two bilateral pairs of actuated joints, as presented in chapter 6, to reduce the rate of trunk extensor muscle fatigue development during prolonged static bending. Since surgical staff report high LBP prevalence levels and associate their LBP with non-neutral, static, or awkward postures, they can be considered a potential use-case of exoskeleton control based on muscle fatigue development. In chapter 5, we objectively assessed the postures of surgical staff and found, interestingly, minimal exposure to non-neutral, static postures. Exposure did not correlate with self-reported LBP or self-reported low-back load. Based on this study, we can therefore not explain the origin of the self-reported LBP in this population. In chapter 7, we discuss the work presented in this thesis, present additional analyses, discuss future perspectives, and draw up conclusions. In summary, this thesis (1) advances active back-support exoskeleton design and control to allow intended movement and provide intelligent support in scenarios involving exposure to low-back loading, (2) progresses our understanding of estimating trunk extensor muscle fatigue development using (high-density) surface electromyography, serving as a stepping stone towards using such a measure in back-support exoskeleton control, (3) provides a basis for an intervention to reduce the rate of trunk extensor muscle fatigue development during static postures that can serve as exoskeleton control strategy, and (4) demonstrates the importance and feasibility of objective, ambulatory postural measurements in complex environments to investigate LBP aetiology and identify practical implementations of exoskeleton support.