Intelligent control for an acrobot (Q1357880)

This paper uses the acrobot as a means of exploring and comparing a variety of intelligent and conventional controllers in a challenging application. An acrobot is an underactuated two-link planar robot that mimics a human acrobot who hangs from a bar and tries to swing up to a perfectly balanced upside-down position with hands still on the bar. The authors develop intelligent controllers for swing-up and balancing of the acrobot. They use classical, fuzzy, and adaptive fuzzy controllers to balance the acrobot in the region near its inverted unstable equilibrium position. They also develop a proportional derivative controller with inner-loop partial feedback linearization, state feedback and fuzzy controller to swing the acrobot up from its stable equilibrium position to the inverted region. In the inverted region, the authors employ a balancing controller to ``capture'' the acrobot and balance it. The authors also develop a genetic algorithm for tuning the balancing and swing-up controllers. The demonstrate how the genetic algorithm can be used to optimize the performance of the controllers.