Assistive Robot Simulator for Multi-Objective Evolutionary Algorithm Application

Z Mohamed,S Khusairi,M A Ayub,M H.M Ramli,M S.B Shaari

doi:10.14419/ijet.v7i4.27.22506

Abstract

This paper presents a new assistive robot simulator for multi-objective optimization application. The main function of the simulator is to simulate the trajectory of the robot arm when it moves from initial to a goal position in optimized manner. A multi-objective evolutionary algorithm (MOEA) is utilized to generate the robot arm motion optimizing three different objective function; optimum time, distance, and high stability. The generated neuron will be selected from the Pareto optimal based on the required objectives function. The robot will intelligently choose the best neuron for a specific task. For example, to move a glass of water required higher stability compare to move an empty mineral water bottle. The simulator will be connected to the real robot to test the performance in real environment. The kinematics, mechatronics and the real robot specification are utilized in the simulator. The performance of the simulator is presented in this paper.

Highlights

A robot simulator is very important for a researcher to simulate path and trajectory of a real robot
The Pareto front has 12 neural controllers and the best neural controller had been chosen from the generated Pareto front, Neural Controller 1 (NC1)
The neural controller shows good performance optimizing all three-objective function, shortest time, shortest distance and high stability. It can be observed from the motion the speed of the robot arm is constant and produce the optimum distance and time (Figure 9)

Summary

Introduction

A robot simulator is very important for a researcher to simulate path and trajectory of a real robot. In other work by [3], a joint trajectory optimized controller for a humanoid robot simulator had been proposed. Ogura had proposed an integration of robot motion environment and the developed dynamics simulator. The simulator shows good performance and have the ability to compute a collision free and optimized trajectory A virtual simulator for a Mitsubishi Movemaster RV-M1 Robot had been proposed by [9]. The proposed simulator try to reduce the human error when working in a poor environment. Other type of robot simulator for robot research had been proposed by [12] with internet control robot simulator, [13] with 3D collision avoidance robot simulator, [14] with wireless robot simulator, [15] with trajectory optimization simulator and humanoid robot soccer simulator by [16]

Methods

Results

Conclusion