Artificial Intelligent Systems and Machine Learning

D. Elayaraja ¹, S. Ramabalan ²

Affiliations
1 Anna University of Technology, Trichy, IN
2 EGS Pillay Engineering College, Nagapattinam, IN

Abstract

Industrial robots are playing an increasingly important role in industry to meet out the demands of automated system and are expected to sense environmental information and also process that information and perform appropriate action for a wide variety of tasks. A major challenge for these robots is that the traditional control techniques generally require an accurate model of the system and its environment. Inaccurate modeling will hamper the mathematical optimization process and have a direct negative effect on their performance. For these reasons, computation intelligence technique is now regularly being employed, particularly evolutionary computation, fuzzy logic and neural computation etc.
Fuzzy controllers have been proven to be good tool for real time processes but the designer has to manually derive the “if-then rules by trail and error. On the other hand, Neural networks perform approximation of the system but cannot interpret the solution. Neurofuzzy concept combines the two approaches in which Neural networks brings the learning capabilities and Knowledge representation from fuzzy logic.
This paper considers the application of Neurofuzzy adaptive techniques in the design and development for Industrial robot PUMA 560. In real world situation, the environment around the robot is ever changing one. So this research work aims to develop a best controller design for online control of PUMA 560 robot.
For validating results various experiments are going to be conducted. The outcome from this research work will strengthen the field of robot controller design using Adaptive Neuro Fuzzy interface system.

Keywords

Puma 560 Robot, Robot Control, ANFIS Controller.

Full Text

     

T. Imran Hussian ¹, M. Kader Khan ¹, N. Indira ², S. Ramabalan ³, C. Balamurugan ⁴

Affiliations
1 Department of Mechanical Engineering, M.A.M. College of Engineering, Tiruchirappalli, IN
2 J.J. College of Engineering & Technology, Tiruchirappalli, IN
3 E.G.S. Pillay Engineering College, Nagapattinam, IN
4 M.A.M College of Engineering, Tiruchirappalli, IN

Abstract

Robot manipulators are programmable mechanical system designed to execute a great variety of tasks in repetitive way.In industrial environment, while productivity increases, cause reduction associated with robotic operation and maintenance can be obtained as a result of decreasing the values of dynamic quantities such as torque and jerk, with respect to a specific task. Furthermore,this procedure allows the execution of various tasks that require maximum system performance. By including obstacles avoidance ability to the robot skills, It is possible to improve the robot versatility, i.e., the robot can be used in a variety of operation conditions. In the present contribution, a study concerning the dynamic characteristics of serial robot manipulators is presented. An optimization strategy that considers the obstacle avoidance ability together with the dynamic performance associated with movement of robot is proposed. It results an optimal path planning strategy for a serial manipulator over time varying constrains in the robot workspace. This is achieved by using multi criteria optimization methods and optimal control techniques (NSGA-II). Numerical simulation results illustrate the interest of the proposed methodology and the present techniques can be useful for the design of robot controllers. A comprehensive user-friendly general-purpose software package has been developed using VC++ to obtain the optimal solutions using the proposed algorithms.

Keywords

Dynamics of Articulated System, Manipulator Optimal Control, Serial Robot Manipulator.

Full Text