Inertial Forces Posture Control for Humanoid Robots Locomotion

In order to evolve in an environment designed for humans, a humanoid robot (or simply humanoid) is supposed capable of performing different motions depending on the given situation. With the walking paradigm in a mature stage, in recent years, the attention of many researchers has passed to more complicated locomotion modes. Some examples are: climbing stairs (Harada et al., 2004), falling down in a less damaging manner (Fujiwara et al., 2004), jumping (Nunez et al. 2005, Sakka et al. 2005) and running (Nagasaka et al., 2004; Kajita et al., 2005), crawling (Kanehiro et al., 2004), etc. If different control strategies are used for each kind of locomotion, the autonomy and/or versatility of the humanoid can be affected by complicating the problem of implementing each control algorithm and switching between them. To treat this issue it has been proposed (Tiong & Romyaldy, 2004) to differentiate two important parts on the humanoid locomotion system: the motion generator (MG) and the posture controller (PoCtr), see Fig. 1. The objective of the former is to define the desired locomotion using some specific parameters and the later will find the angles of all the actuated articulations of the robot such that this motion is achieved. In this work, we will present a new posture controller which finds the angles of all the articulations of the humanoid ( ) which produces the desired generalized inertial forces ( , inertial Ref Q ) and motion of the extremities of the humanoid ( i Ref ), see Fig 1.