Walking Gait Generation with Foot Placement Control for the HOAP-3 Humanoid Robot

While biology-inspired approaches for computing walking gaits are often based on joint oscillators, whenever precise control of feet placement is needed a more specific approach based on explicit footstep control is usually employed. This paper presents a simple hybrid approach based on a gait generator procedure that generates a patterned walking gait based on trajectories to be followed by the feet and the center of mass of the robot. While the trajectories can be modulated according to high-level parameters offering realtime control of the produced gait, the trajectories can also be modified in order to achieve precise foot placements. Robust balance is achieved by integrating a reactive realtime module which adjusts the overall posture of the robot according to the readings of the feet pressure sensors. We apply our gait generation method to the HOAP-3 humanoid platform and several results are presented.