GA-PSO-Based Cooperation Strategy for Two Home Service Robots Playing Ball Maze Game

  • 陳 奕璇

Student thesis: Master's Thesis


This thesis presents a method of robot cooperation for the ball maze game which includes path planning strategy real-time image sharing robot communication and cooperation The ball maze game requires the robots to plan a path from start point to end point in a maze and the robots have to cooperate with each other to move the ball In the process the robots may have only partial image information of the maze so this thesis proposes an image sharing system which allows the robots to do real-time exchange and merging of images For the path planning problem the thesis proposes a GA-PSO algorithm which is based on particle swarm optimization (PSO) and is combined with a mutation mechanism of genetic algorithm (GA) When the particles have a poor fitness value for a long time the probabilities for mutation will increase Compared with other algorithms the proposed GA-PSO algorithm shows a better performance In order to cooperatively accomplish the ball maze game a cooperation method is proposed in which the robots communicate and exchange information with each other In addition the robots can change roles between leader and follower using the information they receive There are two experiments in this thesis In experiment I both robots have the same information so the system assigns roles to the robots In experiment II a white board is placed on the maze so the two robots can only see the maze on their own side In this case the roles of the robots may change during the game The leader robot merges the maze images and finds a collision-free path The experimental results demonstrate the effectiveness of robot cooperation
Date of Award2015 Aug 19
Original languageEnglish
SupervisorTzuu-Hseng S. Li (Supervisor)

Cite this