Quasi-static balancing for biped robot to perform extreme postures using ducted-fan propulsion system

Abstract

Quasi-static balancing is important for enabling humanoid robots to move through extremely rugged terrain, e.g., stepping over a ditch whose width exceeds the robot’s leg length. In this study, to overcome such challenges, an innovative solution was developed, in which external thrust is utilized to maintain the robot’s balance. Initially, a model of the robot’s balance was established for analyzing the factors affecting the balance and the means to maintain it. Subsequently, a new controller combining a thrust controller and a center-of-mass controller was developed to compensate for errors in the thrust output and mass distribution. Finally, a new motion-planning method based on line search regression (LSR) and grid search optimization (GSO) was developed. A series of experiments were conducted using a prototype robot (Jet-HR3), including an error compensation test, an external force disturbance test, a comprehensive motion test, and an active sliding steering test. The results indicated the effectiveness and efficiency of the proposed method. The robot successfully crossed a ditch 695 mm wide, i.e., 147% of the robot’s leg length.