Balancing Control of Robot Gymnast Based on Discrete-Time Linear Quadratic Regulator Technique

摘要

The most important purpose of this paper is to stablise and balance a robot gymnast (Robogymnast) upside down. The Robogymnast is a complex non-linear triple inverted pendulum system. Robogymnast attached to a freely rotating high bar mounted on ball bearings. It mimics the human acrobat, consists of three links and three joints, first joint is passive (unpowered joint) and the rest are active (powered joint). The passive join represents a great challenge to balancing the robot at the upright position. In order to achieve the required performance, the optimal control theory was used. The discrete-time linear quadratic regulator (DLQR) controller was carried out using the linearized mathematical model of the plant which was used to calculate the state feedback control law. Apart from this, the choice of weighting matrices of the DLQR was discussed. The results showed successful stablising and balancing of Robogymnast.