This paper proposes Extending the Selectively Damped Least Squares
method to cater for joint limits avoidance (JLA) when solving the
inverse kinematics of anthropomorphic robotic hands. Here, the SDLS is
considered as the main task allowing to avoid singularity behaviors by
computing for each singular value an appropriate damping factor of a
Jacobian matrix concatenating the proper kinematics of the robotic hand
with an additional task (JLA) projected into the null-space of the main
task. A comparative study of different orders-norm of the objective
function defining the JLA task has been done with the aim to select the
suitable order which is able to overcome the mentioned constraints with
less instability and errors. After an initial tuning to choose the right
parameters, the method was applied to calculate the kinematics of all
fingers’ joints of a robotic hand to fulfill a grasping task. The method
achieves the solution of all fingers together as one complex multibody
system.