Study On The Balance Algorithm Of Humanoid Robot Fast Dynamic Manipulation

At present, leg motion ability of humanoid robot is gradually completed, and the arm manipulation ability develops rapidly with various manipulation patterns springing up. Current researches on arm motion mainly focus on the manipulation itself, which is usually slow manipulation without rigor stipulations on manipulation time. Little attention is paid on the balance maintenance of arm fast manipulation. However, the robot body will be impacted by all-dimensions time-various large inertial impact force and moment when fast dynamic arm manipulation is implemented, therefore the balance maintenance of humanoid robot fast dynamic manipulation is an inevitable and significant research issue.This paper focuses on balance maintaining in the fast arm manipulation of humanoid robot.The content of study is mainly composed of the following points:1) Impact and stability in the fast dynamic arm manipulation of the humanoid robot are analyzed precisely and the intrinsic features of fast dynamic manipulation are extracted out. The fast dynamic arm manipulation will introduce 6-dimensions time-various large inertial impact. Therefore robot manipulation stability of all directions will be examined. By dynamics analysis, the direction, amplitude, characteristics and time-various rule of the impact are preliminarily grasped, thus laid the basis for the difficulty assessment and dimension selection of the balance control. The meaning of stability analysis lies in grasping the change rhythm, tendency and danger region of stability margin in advance, and exhibiting the intrinsic feature of fast dynamic manipulation.2) The specialized resolved momentum control algorithm for robot dual-arm motion is proposed and applied in the balance maintenance problem of robot arm fast dynamic manipulation. The simulation result verified the validity of the proposed balance control algorithm which generates smooth balance motion of auxiliary arm and acquires sufficient ZMP stability margin and remarkable real-time performance.3) The method of joint over-speed phenomenon analysis in view of the real symmetric matrix singularity within matrix generalized inverse algorithm is proposed. This phenomenon frequently emerges in two-dimension situation with improperly chosen dimensions and multi-dimension situation resolved momentum control. The intrinsic reason is found out to be the singularity in matrix generalized inverse, and the severity, universality and inherence of over-speed phenomenon in the multi-dimensions resolved momentum are exhibited.4) The concept and algorithm implementation of humanoid robot full balance is proposed. The method of 6-dimension momentum change rate constrains for full balance control is put forward and applied in humanoid robot arm fast dynamic manipulation. The discretization and convex quadratic programming methods are adopted for optimization constraint control solution. It coordinates balance constraints, joint speed and acceleration constraints of auxiliary arm concurrently. The feasibility of realizing 6 degrees of freedom all-directions robot body balance maintaining by 6-dimensions momentum change rate constraints is verified by simulation. The auxiliary arm balance motion with joint speed and acceleration within limit as well as large angular margin is obtained. It has comprehensively optimal performance of 6-dimension dynamic balance, with large ZMP stability margin and preferable real-time performance.