ANFIS-Based Inverse Kinematics and Forward Dynamics of 3 DOF Serial Manipulator

Study of a serial manipulator involves many non-linear functions and complex sets of equations. The complexity of the problem increases as mobility of the system increases and it becomes a very tedious task to analyse the robotic system especially the inverse kinematics and forward dynamics. This study is aimed to reduce the complexity of analysis by implementing adaptive neuro-fuzzy inference system-based inverse kinematic analysis and forward dynamic analysis. The CAD model of serial manipulators is prepared using SOLIDWORKS to get geometrical and inertial information. The manipulator is also mathematically modelled in MATLAB and its forward kinematics (FK), inverse kinematics (IK), and dynamic analyses are carried out. The ANFIS networks are trained using data generated by a mathematical model, and it is then tested for a new set of data. Finally, the results are validated by comparing with that of the mathematical model.