抄録
The purpose of this study was to propose a methodology which realizes residual error reduction of whole-body CG's acceleration by modifying body specific parameters (e.g. mass, moment of inertia, and position of CG at each segment) for a dynamic contribution analysis of human movements. Since the dynamic contribution analysis, which decomposes generalized acceleration vector of target system into joint torque, gravity, and motion dependent terms, is based on the equation of whole-body motion, a type of analytical formula of the equation of motion was derived by the combination of 1) equations of motion for individual segments and 2) geometric constraint equations of segments connecting to its adjacent segments with the joints. In order to obtain the relationships between the inertia parameters and the whole-body CG's accelerations, a sensitive matrix relating inertia parameter vector to the whole-body CG's acceleration vector was calculated by adding 1% perturbation in nominal value of each inertia parameter. And then by using the sensitive matrix and the error acceleration vector at each time, an error equation with respect to the inertia parameters was derived in matrix form expression. A renewal parameter vector was calculated by the method of least square from the error equation with use of a constraint condition and a step size coefficient in order to deal with the significant nonlinearity of the system. A linked multi-segment simple simulation model, 4-link rigid segment model connecting to the ground via a virtual joint, was used to clarify the efficiency of the proposed method. From the results obtained in the simulation, the proposed method enables us to reduce residual errors of CG acceleration in the dynamic contribution analysis for a whole-body system.
