Planning And Application Of Biped Robot Based On Genetic Algorithm

Biped robot has agile movement direction and speed range. It can adapt to all kinds ofcomplicated ground environment. So it is one of the important research topics in robotresearch areas. Biped walking style is the highest degree of automatic walk style in all walksway. At present the focus of robot is how to make biped robot to walk steadily. When researchthe walking of biped robot, planning technology is one of the main parts. Unlike mobile robot,Biped robot needs to plan all the size changes of joint angle in the robot path. This thesis isbased on the original humanoid robot AFU2010in the laboratory, and it mainly covers thefollowing points:1. Building environment model of biped robot: this paper proposes a method that dividesthe three-dimension environment into three parts: the obstacles layer above robot’s height, theobstacles layer below robot’s moving foot height, and the middle obstacles layer between theabove two sections. The top obstacles will not be considered, the middle layer obstacles can’tbe crossed, and the bottom obstacles for double foot can be crossed. Using this environmentmodeling method can divide biped robot planning into two parts.2. Building kinematics model of biped robot: Firstly, establish the robot body coordinatesmodel. Then establish the reference coordinate system base on the point of support footcontact with the ground. Then establish the local coordinate system of each joint angle.Finally, deduce the coordinate transformation matrix of each local coordinate system in thereference coordinate system. So as to realize the forward kinematics modeling by the robotjoint angle solves the robot connecting rod pose. Then solve the inverse model of the forwardkinematics model through the geometric method. That is realized the inverse kinematicsmodeling by the robot connecting rod pose solves the all joint angle. Then, through the modelofthree-dimensional linear inverted pendulum can get the walking mode of robot.3. Path planning of biped robot: Firstly, Create a group of footprints from the initialposition to the target position randomly in the environment. Then optimize the footprintsthrough the genetic algorithm. So it can realize footprints planning of biped robot. Throughthese footprints can get the foothold of the robot. According to the combination of walkingmode can realize the path planning of biped robot.4. Simulation experiment: Simulate the method of path planning in differentenvironments, and analysis of them. Finally, apply the theory to the robot AFU2010system toprove the feasibility of this method.