Inverse Kinematics of a RSRR HeIse wheel using Neural Networks

The RSRR HeIse wheel is a 2 DOF mechanism that can transform a circular wheel into a hybrid wheel with multiple limbs. Solving the inverse kinematics of this mechanism requires solving a system of non-linear equations with multiple solutions. In order to find an efficient algorithm for on-line computation of the inverse kinematics, in this work we explore the use of artificial neural networks. The analytical solution of forward kinematics is used to generate the data set for the neural networks training. Different structures of networks are tested and their performance in terms of accuracy is discussed. The results of a simulation using the obtained model shows the efficacy of this method.