On the execution of the secondary task of redundant manipulators

In the paper, we compare two different approaches to accelerate the execution of the secondary task of redundant manipulators when a standard task-priority control scheme is utilised. In the task-priority approach, the vector of the desired increment in the secondary task space is projected in the null space of the primary task. This projection may distort the projected vector and the secondary task may not adequately be executed. One approach to avoid the difficulty is to optimise the weight of the pseudoinverse matrix. The problem is that the change of the weight simultaneously influences the execution of the primary task. In the second approach, we directly optimise the null space projection matrix by minimising the difference between the projected and the desired vector of the secondary task increment. In the latter approach, we have the freedom to choose the weight of the pseudoinverse matrix.