ABSTRACT
The inverse kinematics (IK) problem for the six degrees of freedom (DoF) fullarticulated
manipulator is solved either analytically through closed form solutions, or
numerically through iterative algorithms. Analytical solutions are in general, if exist,
yield redundant solutions at an affordable and reliable computational cost. Numerical
solutions that employ iterative algorithms could yield an accurate single convergent
solution that depends upon the picked initial guess, but at relatively higher
computational cost. In this paper, an efficient numerical solution that employs an
innovative implicit iterative algorithm is applied to the decoupled wrist and shoulder
DoF, with the relevant jacobians approximated by first-order finite difference
schemes. The presented numerical solution has been applied to the HUBO
humanoid robot arm to solve for the IK of a single point in the workspace, and also
for trajectory tracking problems. The results for both cases have demonstrated high
accuracy at an appropriate computational cost suitable for real-time applications such
as IK-based walking of humanoid robots.