부산대학교 도서관

We focus on the realization of the hierarchical inverse kinematics as a practical means of a hard real-time motion planning for a humanoid robot. To achieve complex on-line full body manipulation performance, a task priority is established among the tasks to precisely resolve task conflicts and all algorithms are described in recursive equations so as to minimize computation time. Formulating the kinematic problem in a least-square optimization enables us to derive solutions in a unified and consistent framework. The performance is validated through numerical simulations with bearing mind of rescue robot applications and finally, the inverse kinematic solution is found within 0.3msec on average for 19 degrees of freedom control.