Humanoid Walking Robot: Modeling, Inverse Dynamics, and Gain Scheduling Control

This article presents an overview of bipedal walking models by grouping them into two categories: models with concentrated mass and models with distributed mass. As an example of the models with concentrated mass, a mass-spring inverted pendulum model is presented, accompanied by an analysis. As an example of a more complex distributed mass model, a 10-dof walking robot model is developed and analyzed, including the model kinematics, dynamics and controls, with numerical solution simulations for desired joint trajectories, recorded from real human walking cycle data. Kinematic and dynamic analysis is discussed including results for joint torques and ground force necessary to implement the prescribed walking motion. This analysis is accompanied by a comparison with available experimental data. Finally, an inverse plant and tracking error linearization based controller design approach is described, accompanied by results analysis and conclusions about controller performance. The main contribution of this article is presentation of a novel combination of nonlinear gain scheduling with a concentrated mass model for a MIMO bipedal robot system.