A Solution to Inverse Kinematics Problem Using the Concept of Sampling Importance Resampling

The Inverse Kinematics (IK) problem requires solving the non-linear transcendental equations. It is not always possible to obtain a closed form solution. Besides, the existence of a unique solution or multiple solutions is an important issue. This paper therefore tries to solve the IK problem using the Sampling Importance Resampling (SIR) particle filter. Since the particle filter can represent multimodal belief, the proposed approach has the ability to find multiple solutions to the problem. Simulations were performed on a 6 degree of freedom PUMA-like manipulator. Results show that the method exhibits convergence within 3 iterations of filtering. Also, the present approach succeeds in finding multiple solutions, performs well near singularities and requires no initial estimate. The method is computationally efficient and can be applied to many other serial manipulators with multiple degrees of freedom.