GA - SVM and MLP - BBO to estimate Robot Manipulator Joint Angles

231  Abstract—The kinematic of serial manipulators comprises the study of the relations between joint variables and Cartesian variables. We distinguish two problems, commonly referred to as the direct and inverse kinematic problems. The former reduces matrix multiplications, and poses no major problem. The inverse kinematic problems, however, is more challenging, for it involves intensive variable-elimination and nonlinear-equation solving. In this work, we have used Support Vector Machine with Genetic Algorithm and and Multi Layer Perceptron (MLP) with Biogeography-Based Optimization(BBO) to solve the inverse problem on a manipulator arm, to determine its various articulations. The results of simulation are presented to show the validity of approaches suggested above.