Gait Planning And Neural Network Control For Biped Walking Robot

Biped robot, more flexible and close to the features of human walking compared with other mobile robot, has been a focus and important research direction in the field of robotics.And gait planning becomes the base and key in the realization of walking stability, given that the vital role walking stability plays in the research of biped robot. Control technology, the most critical part of robotics, determines the performance of robot. With the self-development biped robot, this paper conducts a study on the gait planning and motion control, which includes the following content:With the mechanical structure of self-development biped robot, this paper introduces the design scheme and equipment selection of driving and perceptual system, moreover, a brief introduction of structure of hardware and software of the whole control system. Based on the introduction of mathematics of robot kinematics, this paper applies D-H method to obtain the forwards kinematics model of biped robot, and the inverse kinematics model was derived by way of algebraic geometry on solving the problem of robot kinematics.Besides, for the problem of gait planning of biped robot, three-step planning is applied in this paper to plan the straight lien walking gait, making use of the polynomial interpolation of the ankle and hip joint trajectory and the inverse kinematics formulation to solve every joint steering angle, and resolves the function expression of every joint trajectory of biped robot to achieve the parametric representation of gait.For achieving the goal of maintaining stability of biped walking, this paper studies the relationship between the ZMP( Zero Moment Point) location and angle of the ankle in the process of human walking, not only putting forward a way to access to real time ZMP through six-dof-force/ torque sensor, but also designing a stable fuzzy controller. ZMP position is away from the edge of support region by way of real time adjustment of the angle of ankle joint, thereby improving the walking stability. At last, it is applied to the walking test of the self-development biped robot. It turns out the walking stability of biped robot performs better comparing with that of ZMP open-loop control method.RBF is applied to the control of biped robot in this paper for better control effects,bringing forward a design of state feedback controller based on RBF, besides, it is been proved that the said controller satisfied Lyapunov stability through formulas. The said controller is applied to biped robot in walking test, and the result shows that the state feedback control based on RBF is of good trajectory tracking ability.