1A2-K08 Developement of Quadratic Programming Solver for Mechanical Computation of Detailed Musculoskeletal System(Digital Human)

Abstract

Robotics computation theory has been applied for the analysis of a human musculoskeletal system in recent days. There are some challenges to accelerate the computation for the detailed analysis of long time motion or the real-time applications assisting rehabilitation or training. In this paper, we developed a quadratic programming solver for inverse dynamics computation of musculoskeletal models. The solver is based on the active-set method. The computation of the cost function and its gradient vector can be represented by the kinematics and dynamics computation of multi-body systems. We also combined the algorisms for large-scale optimization in order to accelerate the computation.