Optimal design of serial link robotic manipulators via genetic algorithms

Abstract

This thesis proposes the creation of a project methodology that optimizes robotic manipulators for specific tasks and subjected to constraints. During the development of this methodology, a in situ case study will be used as a base for suppositions and chooses. This case study is the in situ hard-coating process of the Jirau hydroelectric dams blades. The methodology created uses Thomas Lambs work as a base and involves dividing the optimization problem into two parts: a geometric optimization and a structural one. Both will use differential evolution in conjunction with a multiobjective weight method. The geometric optimization uses the manipulator segments lengths, joint axis and x,y and z coordinates as gens and the multi-objective fitness function uses the length, dexterity, dexterity gradient and work area to maximize the cost. The structural optimization, on the other hand, utilizes the inner diameter of the manipulator segments as gens and optimizes the weight of the structure and deflection of the end effector. Using the case study, a robotic manipulator was optimized and the methodology tested with success.