Genetic Algorithm Based Robot Workcell Layout Optimization

This paper proposes a GA-based optimization method for robot work cell layout. By applying the engineering experience and knowledge of robotic manufacturing system to the representation of geometric and kinematics restrictions, it could efficiently find out satisfying layout solutions that meet the requirements of practical industrial applications. And moreover, it gains a surprising running speed within a reduced searching space. In addition, this algorithm has been embedded in a commercial robot programming and simulation software, and its performance is verified in a die-cast system.The thesis is organized as following:Firstly, it defines the problem of robot work cell layout in industrial applications, describes the objects and goals of the study, and reviews the current optimization algorithms.Secondly, it presents a mathematical model for robots and facilities in the industrial work cell, designs a special data structure to represent the objects and constraints of the problem according to the practical industrial applications. Besides, it also gives a brief introduction to the kinematics model of typical industrial robots, as well as the calibration method for robot operational space. In addition, a serial of criteria is proposed in order to evaluate the layout performance.Thirdly, following an introduction to the principle, convergence and performance of the popular generic algorithms, the thesis proposes our GA-based algorithm specialized for robot work cell layout optimization in industry. According to the conditions that whether the smart operational space could be reduced or not, two sets of rules are introduced in this algorithm. The performance and convergence of the algorithm are analyzed and compared in the paper.Finally, the algorithm is implemented with C#, and as a plug-in, is embedded into ABB's RobotStudio, so that it could be applied to the practical industrial applications. Two extreme cases as well as two practical examples are devised to test the effectiveness and efficiency of the algorithm, whose results prove that it could successfully find out satisfying layout solutions in a short time.