ZMP based feedback control of ankle joint

This research presents a novel numerically efficient method for calculating the Zero Moment Point (ZMP) of a bipedal walking robot, with a perspective of analysing its stability. In this context, a 1-DOF (ankle) structure has been constructed to resemble a robotic leg, with the force sensitive resistors (FSR) attached on its sole. By obtaining the sensory readings, ZMP has been calculated and analysed. The ZMP is calculated for different configurations of our indigenous model, which is then used for checking its stability. ZMP graphs have been generated from this sensory data and also have been classified into STABLE and UNSTABLE classes. The ZMP plot obtained for stable and unstable configurations is supported by a simulation of a HOAP-2 (Humanoid for Open Architecture Platform-2) robot in Webots and an experiment conducted on NaO robot. An another replica of the indigenous model is made but, with a motor attached to its joint. In order to prevent the model from tipping over, a corrective motor action is taken, when the ZMP calculated lies in the UNSTABLE region.