부산대학교 도서관

This article investigates the human postural regulation problem from the dynamical system perspective, which is also applicable for human-like robotics. More precisely, since the dynamical human posture parameters may change caused by varying load or environment abrupt, the semi-Markov jump process is employed to model the human standing postural dynamics with multiple modes. Furthermore, a novel memorized regulation strategy is developed for guaranteeing the stable standing such that the past memory information can be well utilized. In particular, the asynchronous regulation procedure is considered for better describing the human postural model with mismatched jumping modes. By model transformation and stochastic analysis, mode-dependent regulation criteria with state feedback model are established by convex optimization approach, based on which the mode-dependent regulation gains are designed accordingly. Finally, the feasibility and effectiveness of our proposed regulation strategy is verified via an illustrative example of quiet upright standing posture.