Using Dynamic Simulations to Estimate the Feasible Stability Region of Feet-In-Place Balance Recovery for Lower-Limb Exoskeleton Users

Abstract

In recent years, research into the balancing ca-pabilities of lower-limb exoskeletons has increased with hopes of achieving “crutch-less” stance and ambulation. However, achieving upright stability in underactuated bipedal robotics is difficult. Disturbances due to end-user interactions and actuator limitations further complicate any solutions. The current study was therefore aimed at establishing the generalized balancing capabilities of active robotic lower-limb exoskeletons through the use of predictive dynamic simulations. The ability to balance was assessed through the use of the feasible stability region (FSR), which is the region in whole-body center of mass (COM) position-velocity space where it is possible to recover upright balance through termination of the COM velocity. Direct collocation optimal control was used to estimate the baseline FSR for the human-only and human-exoskeleton system under various conditions. Additionally, Pareto optimization was used to establish trade-offs between the FSR and the motor torques that generate the necessary balance strategies, which determine the FSR. In general, our results indicated that baseline human-only and human-exoskeleton systems share similar balancing capabilities in terms of the FSR, regardless of the device's end-user mobility; however, features of the exoskeleton like high joint-level impedance and a shifted center of mass have detrimental impacts to the overall FSR size. Results from the Pareto optimization suggest that the full FSR can be nearly reached with a fraction of the required motor torques, thus protecting both the device and user. Future work will expand the current analyses to stepping strategies and control-design implementation in the Technaid Exo-H3.