Relocating actuators towards a base frame does not improve weight and inertia characteristics in our upper-extremity exoskeleton

In rehabilitation exoskeletons, alignment of robot and human joints is essential for comfort and performance. Most exoskeletons have their actuators directly at the DOF they control, despite performance being assisted by low movement inertias. For our new upper-extremity exoskeleton, we explored using a parallel robot for the auto alignment of the shoulder axes and three serial links to drive the rotations. We hypothesized that weight and inertia can be reduced by relocating the actuators to the non-moving base of the device, as others have done before. To investigate if this is indeed beneficial, we evaluated all possible topologies for placing motors and gearboxes and are here reporting on the best candidates. We explored several drive trains that combine coaxial axes and angular transmissions. As bevel gears show backlash, are high in weight and large in size, three alternative angular transmissions were investigated. After combining the new angular transmissions with the possible topologies and evaluating the 16 resulting combinations, we concluded that for our new, high-performance exoskeleton, the most optimal topology still is one with the motors and gearboxes placed directly at the joint. The hypothesis therefore was disproven for our usage scenarios.