Piece-Wise Kinematic Optimization of Serial Manipulators with a Pilot Study in Exoskeleton Hip Redesign

Abstract Kinematic optimization is a critical element of the mechanical design of robotic manipulators, especially in the presence of kinematic constraints such as those found in a battery-powered whole-body exoskeleton, where the human operator must be safely followed but not impeded, all while maximizing joint efficiency. In the quest for an optimal robot, especially one of such complexity, effort and time can be reduced by focusing on a particular subsection. A method is presented for kinematic optimization early in the (re)design of a subsection of an existing or proposed serial robotic system. The subsection to be optimized is isolated and then boundary loads and speeds are enforced from data of whole-system operation. As a case study, the energetic optimization of the hip joints of a full-body powered occupational exoskeleton is demonstrated based upon characteristic tasks performed by a user. A cost function is defined that minimizes torque as proxy for consumed energy, which is demonstrated to save 55% computational time, while avoiding poor manipulability and self-collision. The resulting hip kinematics reduced energy consumed by the hip joints when performing core tasks by 13% without worsening manipulability, which is important for ensuring robustness to unexpected conditions. This benefit comes from minor adjustments to design only and without the complexity, weight, or bulk of additional elements.