Research On Motion Planning And Optimization Algorithm Of Humanoid Manipulator

Humanoid manipulator was designed according to the bionic structure of human arm.It has many characteristics such as more joint degrees of freedom,compact structure and more flexible movement.Specifically,the motion planning of humanoid robots has been a hot topic in the field of robotics research,and it has been widely concerned by many researchers domestically and overseas.Motion planning of humanoid manipulator mainly includes the consist of two parts.Firstly,trajectory planning is planning a motion trajectory from the initial state to the target state in joint space or Cartesian space.Secondly,the inverse kinematics solution is mainly the relationship between the motion state of the end effector in Cartesian space and the joint angle vector in joint space.This thesis's research object is based on the humanoid manipulator with Series Elastic Actuator joints as research object,.to study the motion planning and optimization algorithms of the humanoid manipulator.We focus on trajectory planning and inverse kinematics of humanoid robots,especially the analytical solutions based on quadratic programming.The main contributions are as follows:1.An end effector smooth motion planning algorithm based on kinematic limits is studied.The method takes the shortest movement time as the target,mainly aimed at the planning of the end effector trajectory,including point-to-point and transition segment trajectory planning.And we studies a method of time-phase synchronization in multi-dimension.2.A kind of inverse kinematics method for solving humanoid manipulator with redundant degrees of freedom is studied.Based on this method,a new optimization method of joint angle is proposed.Taking the shortest movement distance of the manipulator as the optimization objective,joint angle is optimized by using off-line or on-line discrete optimization.It can make full use of the redundancy characteristics of the humanoid manipulator and increase its flexibility of the motion.3.A kinematic method based on quadratic programming is presented,which transforms the relationship between the end position and joint angle of manipulator to velocity and jerk level.The solution to the problem has solutions based on linear variational inequalities and numerical solutions.In addition,a planning scheme of zero-space self-motion based on this method is studied,a planning scheme of humanoid manipulator obstacle avoidance at velocity level and motion planning of humanoid manipulator at jerk level are proposed based on this method.