Optimization and Evaluation of Spinal Exoskeleton Design Concepts using Optimal Control

Exoskeletons for the lower back are promising tools to support workers during heavy lifting tasks. Their development process faces several challenges. It is still not known which criteria the support must meet to prevent low-back pain and how users of different body stature and execution of lifting movements influence it. Thus, studying these factors needs an extensive testing on the human body and every considered design concept needs already a sophisticated prototype that subjects can wear for several hours. To overcome this issue, we propose a method using multibody dynamics and optimal control to optimize the design of an existing prototype (PO) as well as evaluate a new concept (DC) that incorporates motors at the hip joint. A dynamic model of the prototype with matching torque generation was developed, which also takes an approximation of possible misalignment into account. The human-robot interaction is simulated in an all-at-once approach that allows to calculate the muscle activity of the user required in addition to the exoskeleton support to reproduce recorded lifting motions. By minimizing the users' muscle activity, parameters describing the characteristics of the passive elements and, in case of DC, motor torque profiles are optimized. Compared to the initial setup, a significant improvement in exoskeletal support was achieved across all subjects in both cases while contact forces remained within prescribed limits to ensure a comfortable usage of the device. DC provides less support than PO but better control of the human-robot interaction.