抄録
We have already formulated a human three-joint arm's optimal control model with a freezing-like mechanism in its hand joint. There are the following two discrepancies in reproducing performance of reaching movements between the three-joint model and conventional two-joint models: (1) for the energy minimization criteria, whereas the torque-sustaining energy of the three-joint model is not involved in reproducing reaching movements, that of the two-joint model is involved; (2) for the torque-change minimization criterion, whereas the three-joint model can not reproduce reaching movement characteristics, the two-joint model can do. In order to clarify the reasons for the above discrepancies, we measure human two-joint and three-joint arm's reaching movements and examine the both models' reproducing performance. Consequently, the following results are obtained regardless of subjects and the number of the model's joints: (1) for the energy minimization criteria, it is highly possible that the torque-sustaining energy is not involved in reproducing these measured movements; (2) the torque-change minimization criterion can improve the reproducing performance by adjusting the physical parameters such as the coefficients of the model's joint viscosity or the moment of inertia of the model's links. These results mean that the discrepancy in the energy minimization criteria can be resolved, where that in the torque-change minimization criterion can be caused by inappropriate physical parameter values. Therefore, they suggest that human arm system executes reaching movements based on a criterion that it moves its joint angle as little as possible.
