Solving inverse kinematics trajectory tracking of planar manipulator using neural network

Abstract

Inverse kinematics is the context of controlling multiple joints of a robot arm with the method of estimating the joint angles from the given end-effector coordinates. In this paper, different architecture of neural network Backpropagation Levenberg Marquardt was proposed and analyzed to solve inverse kinematics trajectory tracking in a trainer planar three-link manipulator. The result shows the amount of neuron(s) in hidden layer affected to the result of the solution. Improving the performance of the neural network can be addressed in the future work.