This paper proposes a novel method for real-time trajectory generation of a seven-link planar biped robot. Individual jointtrajectories are generated by specifying only two key parameters, walking speed and step length. The proposed method combines several methods and concepts including kinematics, dynamics,trajectorygeneration,experimentaldesign,Takagi–Sugunofuzzysystems,centralpatterngenerator (CPG), ZMP criterion and a dynamic balance method. A fuzzy relationship between walking and CPG parameters is learned using experimental design methodology and T–S fuzzy systems. A method based on the Fourier series is used to tune parameters ofthe CPG. The proposed method allows making online changes to step length and walking speed while ensuring robot’s dynamic balance. A multibody simulation package is selected and the effectiveness ofthemethod is illustrated using several examples. It is also shown that changes in joint angles, as the result of online changes to the walking parameters, occur in a smooth and continuous manner.

Keywords