Inverse kinematics of an equal length links planar hyper redundant manipulator using neural networks

An iterative method using the neural networks to solve the inverse kinematics problem for equal length links redundant manipulators is presented in this paper. The training phase, calculating the neural networks weights, is accomplished for a new proposed geometrical method to solve the problem of multi-solution caused by redundancy. The use of this geometrical method results in one solution among the infinite solutions of the inverse kinematics of the redundant manipulators. This method is very effective for avoiding the singularity problem because it guarantees that there is no lining up for two or more links. Another advantage for this method is that the angles between the links will be set between two maximum and minimum values. This means that the end-effecter can reach any point on the desired path and the angles between the links will not be less than the minimum limit or more than the maximum limit, which makes this method effective for joint limits. To demonstrate the effectiveness of this proposed method, experiments were conducted on an 8 links hyper redundant manipulator in this paper. In addition, the workspace of the manipulator is calculated for this proposed method