Online Trajectory Replanning for Biped Robots Based on Foot Step Position Modification

Trajectory generation of many successful biped robots is based on the zero moment point (ZMP) criterion. Combining this criterion with a simplified model of the robot dynamics allows for an elegant formulation of the equations of motion of the robot. As solving these equations as an initial value problem does generally not lead to trajectories that allow for stable walking, the desired robot state at the end of the planning horizon has to be considered as well. The boundary value problem that arises is ill-posed and can only be solved in the general case if additional degrees of freedom are introduced. In this paper we give an overview of the real-time trajectory generation system of the humanoid robot Lola that uses the ZMP criterion and introduces additional degrees of freedom in the ZMP trajectory. We propose a modification of the foot step positions to reduce the amount of ZMP trajectory modification necessary.