Topology optimization of industrial robots for system-level stiffness maximization by using part-level metamodels

This investigation presents a topology optimization method for the design of lightweight serial robots for industrial applications such as welding robots and painting robots. It might be numerically efficient to perform topology optimization of a robot structure by dividing the problem into part-level optimization problems. However, the robot structure whose parts are separately optimized is not necessarily the optimized structure in the system level. For example, a robot whose parts are separately designed to have maximum stiffness-to-mass ratio cannot have the maximum stiffness in the system level. This is because it is impossible to know in the stage of the problem formulation how the total mass should be divided into each part to have maximized system stiffness. To deal with this, a metamodel relating the stiffness and the mass usage is constructed in each part-level optimization problem. The proper division of a mass in the part level is determined by solving the system-level optimization problem which is formulated by using the part-level metamodels. Optimized robot structures obtained by the proposed approach are shown to have performances close to system-level optimized ones in test problems with two- and three-dimensional static and dynamic cases. Based on the proposed idea, topology optimization of a painting robot is performed; a base frame, a lower frame and an upper frame of the robot are optimized to lower the maximum system strain energy during the motion.