Research On Kinematic Control Of Manipulator With Multiple Constraints

With the rapid development of the robot industry,robotic manipulator has been widely used in manufacturing factories,and spread to the application of medical field and family services.There are different kinds of manipulators,with prismatic or revolute joints,two to seven joints,and dif-ferent velocity and load capacity.It is crucial to develop kinematic control methods suitable for those manipulators.On the other hand,several industrial field buses have met the requirement of manipulator's real-time kinematic control.And the digital processor could finish some complex al-gorithms at each control period.Thus,real-time control is adopted instead of the off-line planning,and the robustness of kinematic control method becomes more important.Moreover,some ma-nipulators work in the complex and dynamic environment,and the ability of dealing with multiple tasks and constraints is necessary.This research focuses on manipulator kinematic control under attitude pointing,joint limit constraints,task-priority hierarchy,and multiple constraints.Theoretical research with simulation and experimental verification is carried out.All those methods are tested on a seven-degree-of-freedom pingpong manipulator,and some of the results are adopted in the industrial robot soft-ware as a part of the motion planning algorithm.The major research works and contributions of this thesis include:1)A general framework for the kinematic control of manipulator with attitude pointing task is developed in this thesis.Attitude pointing task keeps the direction of end-effector instead of its attitude,which is widely used at applications of manipulator.The nonsingular unit quaternion is adopted in the manipulator kinematic control.The expression of attitude pointing error using quaternion is introduced,and Lyapunov function is constructed to prove the global stability of the proposed method used in the closed-loop inverse kinematic control.By using attitude pointing control,the manipulator would get an extra degree of freedom compared to the ordinary attitude control.In return,the manipulator would have better dexterity and manipulability.The ability of avoiding joint limits and obstacles is also improved.2)Most constraints can be converted into a virtual joint-limit problem by the general weighted least-norm method reported recently.However,a chattering of weighted factors might occur in the presence of disturbances.In this thesis,in order to improve the robustness to disturbances,a clamping term forcing joints away from their limits is added to the solution of inverse kinematics,termed as the clamping weighted least-norm method.It is proved that the joint-limit constraint is always guaranteed.The proposed method is applicable to both non-redundant and redundant manipulators;and the accuracy of main task is sacrificed only when there is confliction between main task and joint-limit constraint.3)Task-priority based kinematic control of robotic manipulator is an effective technique to deal with multiple tasks.In order to address complex tasks,the insertion of tasks is necessary,such that the manipulator will focus on the important tasks.However,the task-priority based control is inherently discontinuous when the insertion of tasks occurs.In this thesis,a new form of calculating null space projection is proposed,by which,the smooth projection operator and the continuous inverse operator based on activeness of tasks is given.Such that task-priority control still keeps the continuity of joint velocity during the insertion of tasks.4)The general weighted least norm method introduces the concept of virtual joints to cope with the multiple constraints that are regarded as virtual joints with joint-limits.With the perfect perfor-mance for joint-limit avoidances,the constraints could be handled well with kinematic control of redundant manipulators.However,the singularity of the map from joints to virtual joints has not been well addressed,and the trick to combine some constraints into one constraint looks artificial,which is required when the number of constraints is larger than the degree of freedom.In this paper,an improved formulation of virtual joints is presented to overcome the above two drawbacks so as to complete the virtual joint concept.