Fortunately, we got answers from Mathworks staff. From the conclusion, it seems that the way of connecting the above figures is wrong. I changed the configuration and procedure of the robot simulation model. The procedure is as follows. First, calculate the angular velocity and torque of the joint using an inverse kinematic model. Using this as a target value, we decided to evaluate how much the hand position can be reproduced by giving torque and angular velocity to the dynamic simulator.
The answers from the mathworks staff were automatically translated from Japanese to English and written below.
Simscape Multibody and Simscape are linked by power. Simscape Multibody's hand position is given to the model consisting of Simscape model, and the torque of Simscape is obtained by inverse dynamics. Furthermore, both the speed source and torque source are applied to the Simscape circuit. It is to input and calculate.
In conclusion, it is difficult, and I think that it is difficult to confirm the coincidence of dynamics calculation and inverse dynamics calculation for the linked model.
For example, can you model only Simscape Multibody?
Actually, the Sismcape part is deleted, the torque of the Revolute joint is obtained by the reverse dynamics calculation, then the forced displacement of the hand is deleted by the dynamics calculation, and the torque obtained by the reverse dynamics calculation is obtained by the Revolute joint. When entered as the torque, the behavior was almost the same.
