Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
HUMANOID WALKING FOR A ROBOT WITH CURVED FEET
Submission Author:
Caroline Silva , RJ
Co-Authors:
Caroline Silva, Marcos Maximo, Luiz Carlos Góes
Presenter: Caroline Silva
doi://10.26678/ABCM.COBEM2021.COB2021-1080
Abstract
In this paper, we propose to develop a walk in a humanoid robot, which contemplates the vertical oscillation of the center of mass (CoM) together with curved robotic feet. Humanoid robots are fascinating because they have a human-like appearance. But in reality, their behavior differs from humans. The battery cannot last for a long time, limiting the use of the robot. For example, the walking of a state-of-the-art Asimo robot uses more than ten times the energy of a human. This difference is because robotic movements are complex from the point of view of control. Due to nonlinearity, sub-actuation, and high dimensionality, representing a challenge to the latest generation control techniques. In the biomechanical literature, some studies show reasons for the human gait to be energetically economic. One of these reasons is that the curvature of the human foot provides a bearing similar to a wheel, which decreases the cost of walking. Our contribution is to adapt the walk for curved feet, which considers vertical oscillation of the CoM. In this work, we implement the Preview Control of Zero-Moment Point technique to plan the trajectory of the CoM. It is a method of bipedal walk control based on Model Predictive Control (MPC). State-of-the-art robots, such as the ones designed by Boston Dynamics use MPC for walking. Due to the many degrees of freedom of the humanoid robot, we apply the 3D Linear Inverted Pendulum Model (3D-LIPM), to the dynamic model of the Preview Control of Zero-Moment Point technique. For the curved robotic foot, we propose a geometry formed by three parts: a cylindrical sector corresponding to the heel, a parallelepiped for the sole, and another cylindrical sector representing the toes. Subsequently, inverse kinematics (IK) finds the angles of all joints. We validate our approach by implementing the proposed control technique described above with the parameters corresponding to our custom-made robot Chape. Our results showed a stable gait, encompassing human-like mechanisms in a robot.
Keywords
humanoid robots, MPC, Walking, Robotic Foot
