Push recovery for NAO humanoid robot

Abstract

This paper presents the ankle, hip, and ankle-hip strategies in frontal plane for NAO humanoid robot. A humanoid is an unstable robotic mechanical system and the main and primary task of each humanoid is to maintain its balance. Moreover, NAO humanoid robot has 25 degrees-of-freedom with brushless DC actuators. The main challenges in the context of balancing of NAO humanoid are the high complexity of the robot and applying joint torque control approaches for NAO humanoid which is a position controlled robot. In this paper, for recovery and prediction the robot from external push, as the main contribution, the virtual leg model is proposed. The control objective is absorbing the external push and then recover the robot to its original configuration. PD controller is used to achieve the control objective. The performance and usage of these strategies and the model is validated by Webots. For each case study, an impulsive force is applied to NAO's torso with various magnitudes. The performance and validation of the proposed push recovery schemes are verified in the simulation. Moreover, the same results are obtained from the practical implementation of these strategies on NAO H25 humanoid robot.