We extend the range of inverse kinematics by integrating mass distribution information to embody the position control of the center of gravity of any articulated figure in single support (open tree structure). The underlying control architecture of main and secondary behaviors (or tasks) is used to associate this new approach to other behaviors. We evaluate the use of these techniques for motion control and we present cascaded control as a way of hierarchically combining balance and other goal oriented behaviors. The combined quality of providing realistic motions in real time significantly improves the potential interaction with the human model in virtual reality applications and, more generally, enhances the posture and motion design process for complex articulated figures