An increasing range of engineering applications require soft grippers, which use compliant mechanisms instead of stiff components to achieve grasping action, have high conformability and exert gentle contact with target objects compared to traditional grippers. In this study, a three-fingered gripper is first designed based on a notched self-folding mechanism actuated using an electrostrictive PVDF-based terpolymer. Then the design optimization problem is formulated, where the design objectives are to maximize the free deflection Δfree and the blocked force Fb. A computationally efficient two-stage design optimization procedure is proposed and successfully applied in the gripper design. NSGA-II is adopted as the optimization algorithm for its capacity to deal with multi-objective optimization problems and to find the global optima with high design variables and large design domains. In stage one, computationally less expensive analytical models are developed based on Bernoulli-Euler beam theory and Castigliano’s theorem to calculate Δfree and Fb. Utility function is applied to determine the best design in the last generation of stage one. In stage two, 3D FEA models are developed, using the dimensions determined by the best design from stage one, to investigate effect of the shape of segment surfaces on the design objectives. Overall, the proposed two-stage optimization procedure is successfully applied in the actuator design and shows the potential to solve a wide range of structural optimization problems.