1 Robust multi-objective design optimization of the 3-UPU TPM based on the GA-Krawczyk method S. EL HRAIECH 1 , A.H. CHEBBI 1 , Z. AFFI 1 , L. ROMDHANE 2 1 LGM, National Engineering School of Monastir, University of Monastir, Tunisia e-mail: safa_el_hraiech@hotmail.fr ahmed.h.chebbi@gmail.com zouhaier.affi@gmail.com 2 Department of Mechanical Engineering, American University of Sharjah, UAE lromdhane@aus.edu Abstract. This paper deals with the robust design optimization of the 3-UPU translational parallel ma- nipulator. An approach, that regroups the genetic algorithm multi-objective optimization and the Krawczyk operator (GAMOK), is used to represent the optimal design vector of parameters and their uncertainties. This optimization leads to minimize the position error and relax the parameters intervals of tolerance. Based on this GAMOK algorithm, the designer can pick out the optimal design vector ac- cording to the desired accuracy in the workspace of the manipulator. Key words: Interval analysis, Krawczyk operator, Genetic algorithm, uncertainties, optimization. 1 Introduction Parallel manipulators have many advantages such as, greater rigidity, higher stiff- ness and essentially higher accuracy compared to serial ones. There are several types of parallel manipulators; the translational robot, the rotational robot and the mixed ones [9, 10]. The position error of parallel manipulator caused by design parameters uncertainties cannot be neglected. Therefore, it is quite important to optimize the design parameters and their uncertainties as function of the robot per- formances. Several optimization methods have been used. Genetic algorithm (GA) is an evolutionary algorithm inspired from natural evolution, used to solve optimi- zation problems [1]. The main advantages of the genetic algorithm are the capabil- ity to escape local optima and its powerful searching ability. Laribi et al. proposed Author's version