There are two broad classes of solutions to inverse kinematics problems:analytic and numerical. Numerical solutions are the most general and are able to handle sophisticated constraints with complex articulated figures. Welman[1989] presented a good suuvey of computational techniques. Though powerful ,numerical IK algorithms suffer from many drawbacks.Since they require solving systems of nonlinear equations, the solutions are computationally expensive. Convergence is difficult to guarantee ,and as the IK problem is inherently ill-conditioned[1990] ,the algorithms can be unstable.