Multi-objective optimization to improve the performance of Delta Robot for a predefined workspace

Due to conflictions between different purposes and objectives of the industrial robots, a multi-objective optimization is required to find the non-dominant answers to choose from. In this article, we have introduced a new optimization objective which involves the robot’s high acceleration, actuator effort reduction; while avoiding the singularities which is applicable on any robot with an approximate known inverse dynamic. The cost function is defined by designing a computed torque control on robot. Our objectives are defined as: 1. minimizing the root mean square (RMS) power needed in motors for movements with the maximum acceleration of the end effector in different directions (cost/effort reduction), 2. Minimizing the mean Jacobian condition number in a predefined workspace (kinematic performance improvement) and 3. Minimizing the mass matrix condition number in a predefined workspace (dynamic performance improvement). The multi-objective problem is numerically solved with Genetic algorithm SPEA-II and non-dominant solutions known as Pareto fronts are resulted. Geometrical design variables are chosen from Pareto fronts by three different MCDM methods. Eventually, cost functions of the designed robot are compared with a Fanuc M-2iA/3S model.