Humanoid robots have a similar appearance to humans with functions of mobile,perceptive and operation.This category of robots not only can accomplish operations independently,but also can cooperate with human beings to complete tasks.In fact,humanoid robots operations mainly be executed by their high degree of freedom armtorso system,after the mobile platform arrives the designated location.However,armtorso system has characteristic of high redundancy,the research emphasis on armtorso system is how to obtain a high dexterity trajectory,to avoid task failures caused by additional constraints during operations.Moreover,flexible components,such as harmonic reducers,are contained in arm and torso joint transmission chain,which gives rise to inevitable vibration of humanoid robots in motion.Whereas,excessive vibration will bring about task failures.And hence,vibration during humanoid robots motion must be suppressed.For the sake of completing the entire design of humanoid robot,the design requirements of humanoid torso are first determined in this research,via analyzes the parameters of humanoid arm and humanoid head,the top-down design sequence and the operation task.Afterwards,degree of freedom and geometric configuration of humanoid torso are confirmed through the movement analysis of human torso.After the torso joint design of mechanical structure,internal sensor syste,electrical system and safeguard,control parameters of torso joint are calibrated by parameter identification experiments.Finally,other subsystems of humanoid robot are assembled to the torso and the humanoid robot system is completed.Besides,main dimensions of humanoid robot are compared with the adult dimensions,the results indicate that overall dimensions of humanoid robot are relatively coordinated.Obviously,via compares the volume of humanoid robot workspace with rigid torso and dynamic torso,the workspace of humanoid robot is evident expanded with a dynamic torso.The performance parameters of torso joint indicate that the desi gned torso joint has advantages of high reduction ratio,large output torque,and short axial dimensions,.Meanwhile,the torso joint also integrate a variety of sensors,as well as has characteristics of electromechanical integration.Range of workspace and operation dexterity in workspace are significant indicators to evaluate performance of humanoid arm.In this paper,pose dexterity index,whcih is defined by range of arm angle and inverse kinematics solution,is used for describing the dexterity of manipulator in specified grasping direction.Compares with traditional dexterity index,pose dexterity index considers probability of manipulator reaching target position in specified direction,simultaneous takes range of feasible arm angle in consideration,to achieve more comprehensively description on manipulator operational dexterity.In addition,analytical method is used for inverse kinematics solution,which can avoid the inaccuracy of pose dexterity index caused by the inverse kinematics solution precision,and also can speed up the trajectory optimization via store the feasible arm angle range or the optimal arm angle.Meanwhile,position dexterity index is defined to describe the operational dexterity of humanoid arm in different workspace areas.Compares with traditional dexterity index,success rate of operation task in high dexterity workspace area generated by the proposed position dexterity index is similar with the reachability,but significantly better than density,manipulability and condition number index.Especially,while accomplish operational tasks with arm angle constraints,the position dexterity index is obviously superiority to traditional dexterity index.Indeed,humanoid robot system has an arm-torso system with high degree of freedoms.Thus,in order to achieve high dexterity motion planning of arm-torso system,a high dexterity workspace area of humanoid arm is established based on the position dexterity index,and torso joint angle is optimized by using distribution of the high dexterity workspace area.Subsequently,the optimal operation position on operation platfrom is obtained in accordance with workspace position dexterity index,as well as,this optimal operation position is defined as target point of the operation,to convenient for achieving complex operations on target.Lastly,in order to achieve a high dexterity of grasping posture,target grasping orientation is optimized by pose dexterity index.After obtaining the key poses and corresponding joint angles for the operation task,spline curve is used for achieving the smooth connection of corresponding joint angles,and the driven joint trajectory is generated by discrete interpolation of joint spline trajectory,thereby to realize a dexterous operation motion planning of the humanoid robot.For operation tasks which is performed repeatedly,an improved particle swarm optimization algorithm is put forward,to achieve vibration suppression with multiple constrained conditions based on trajectory optimization.In this process,residual elastic potential energy is used as object function,and through constraining the joint space control point position(or/and)the Cartesian space control point position,the trajectory optimization of humanoid arm under constrained conditions is completed,thereby joint trajectory with minimum residual vibration is acquired to achieve joint vibration suppression.In addition,for real-time operation tasks,a real-time trajectory modify method based on centroid deceleration of humanoid robot,is proposed to achieve vibration suppression.Firstly,occurrence time of vibration and state of the robot motion are obtainded with the changes of joint torque in motion.Secondly,initial joint trajectory is modified to lower the speed of robot’s center of gravity,thereby reduce the increasing trend of vibration to realize vibration suppression. |