Realistic dynamic posture prediction of humanoid robot: Manual lifting task simulation

Abstract

A well known question mooted in biomechanics is how the central nerves system manages the body posture during various tasks. A 5DOF biomechatronical model of human body subjected to simulate the manual lifting task of humanoid robot. Simulation process is based on optimization approach named predictive dynamics using inverse dynamics. An objective function in term of ankle torques during lifting time, subjected to be minimized. It assumed that CNS considered this function to perform lifting motion balanced. In the other optimization-based simulations, balancing motion was guaranteed by a nonlinear inequality constraint which restricts the total moment arm of the links to an upper and lower boundary. In this method there is no need to use this constraint. Result shows that the motion is performed balanced. According to the comparison the results with the experimental data, the body posture of humanoid robots, predicted as similar as actual human posture. © Springer-Verlag Berlin Heidelberg 2012.