A new method to design fuzzy controller for unmanned autonomous forklift

This paper proposes a new method called Lyapunov-Conjugate Gradient Method (LCGM) to design fuzzy controller for unmanned autonomous forklift. The kinematic of a spin-turn mechanism is analyzed first, and then the design parameters are obtained. Based on the kinematic and the fuzzy Lyapunov synthesis proposed by Margaliot, we introduce the error weights coefficients to Lyapunov Function (LF) and derive the stable conditions. The fuzzy rule-bases are given. Next we design and optimize parameters of the Gaussian membership function by the Conjugate Gradient method. The smoother control surfaces are analyzed. Experiment results show that the type-1 controller designed by LCGM has a better performance in stability, smooth and robustness than commonly used conventional fuzzy-PID controller. Its steady lateral error and orientation error can be less than 10mm and 1 degree, respectively. A lot of unmanned forklifts that their controllers are designed by LCGM have been used in automatic plants.