Proposed IK algorithm and AI controller for TIP position robotic Manipulator.

number: 
1274
English
Degree: 
Author: 
Baker Basil Ali Al-Bahri
Supervisor: 
Dr. Mohammed Z. Al-Faiz
Dr. Mazin Z. Othman
Dr.Safaa S. Mahdi.
year: 
2005

Abstract : The aim of this thesis can be divided into two main parts. In the first part an algorithm to solve the inverse kinematics problem of a complex structure robotic manipulator based on the rotation vector concept is presented. Solution of the inverse kinematics problem of the robotic manipulator represents one of the problems in the field of the robotic systems. In the proposed algorithm the orientation of the gripper is described with the aid of the rotation vectors too. The proposed algorithm to solve the inverse kinematics problem gives all the possible sets of solutions of the inverse kinematics of the manipulator. The mathematical computation and complexity in the proposed algorithm are less than that in the geometrical or numerical methods. The proposed algorithm is used to solve the inverse kinematics problem of the MA2000 robotic manipulator, which gives four sets of possible solutions of the MA2000 robotic manipulator. The MA2000 is a six degree of freedom (6DOF) complex structure robotic manipulator with pneumatic gripper as the end-effector. The forward kinematics problem of the MA2000 is solved with the aid of the Denavi_Hartenberg (D-H) method. The second part of this thesis represents the design and implementation of a multilayer artificial intelligent controller for tip position control of a robotic manipulator (MA2000) based on the fuzzy logic and neural networks algorithms. Three types of AI controller schemes namely fuzzy logic controller (FLC), inverse neural network controller (INNC), and hybrid fuzzy and neural controller (HFNC) are designed and tested practically in the real time operation of the MA2000 robotic manipulator. The FLC gives fast and smooth transient response combines sometime with steady state error; while system response with the INNC has bad transient response but zero steady state error. The HFNC takes the good features of the FLC in the transient response and the good steady state features of the INNC. Practically, it is found that, for small step change in the system response the INNC gives the better performance than FLC. Each joint of the six joints of the MA2000 robotic manipulator is treated as a single input single output system. Thus six separated AI controllers are used to control the operation of the MA2000 robotic manipulator. All the system software is written with the aid of the MATLAB package, which represents the host of all system software and algorithms. In order to combine the operation of the MA2000 robotic manipulator with MATLAB package, a certain dynamically link library (DLL) is developed and used to transfer data between the MATLAB and system hardware. Also a hardware system (physical layer) of the system is designed and implemented to connect the system (MA2000 robotic manipulator) to the PC (1.1 GHz) via ISA bus. A graphical user interface under the MATLAB is deigned and implemented for the real time operation of the MA2000 robotic manipulator.