Dynamic modeling and system identification for the lower body of a social robot with flexible joints

This paper investigates modeling and identification for the lower body of a social robot, Nancy, with flexible transmission. The flexible-joint and motor parameters of the rigid-link flexible-joint (RLFJ) model and the inertial parameters of the rigid model are identified in using the developed linear dynamic parametric model and the Weighted Least-Squares (WLS) technique. The comparison of these inertial parameters are examined identified by different trajectories. The identifiability of the inertial parameters is analyzed. The optimal excitation trajectories are produced with two optimal criteria for the RLFJ and rigid robot model. The performance of the Least-Squares (LS) estimator and WLS estimator is verified by numerical simulation for the RLFJ model and rigid model, respectively. The validity of the identified model is inspected in tracking control ability.