Humanoid Robots Staircase Stepping Stabilization using Novel Previewed Fuzzy Granular Controller

摘要

Humanoid robot staircase stepping requires a zero moment point at each step during climbing. A novel fuzzy controller based on preview evaluates the system by taking a double inverted pendulum movement about the cart as a prototype for the leg motion of the humanoid under the impact of time delay using lookup tables. The fuzzy controller with granular computing brings the pendulum to the upright equilibrium position and then a LQR controller smoothens the final position about the unstable equilibrium position. The advantage of this method when compared with other existing methods is the computation of the system performance under the influence of time delays which is induced into the system state space matrix model. The performance measure is considered in terms of settling time, peak overshoot and offset which predominantly indicate that developing a controller based on fuzzy logic with granular computing and with preview will yield better control which is energy efficient.