Targeting hydrocarbon-rich zones and locating smart wells optimally are critical to oil and gas field development. With model uncertainty being one of the main considerations in well placement, there are several challenges associated with the statistical inference from objective distributions which are normally unknown a priori and the estimation may become computationally intensive. To overcome these challenges, we use a robust multi-objective optimization approach to handle model uncertainty and propose an efficient polynomial-based technique to accelerate optimization process. To handle model uncertainty, we implement an evolutionary multi-objective optimization algorithm (NSGA-II) with a mean-variance approach to seek robust solutions for well placement, where the robustness is considered for the uncertainty of formation characteristics. To assist the optimization process, we apply a polynomial-based sparse grid interpolation approach coupled with the multi-objective optimization algorithm. This workflow is demonstrated on a reservoir case study where the results indicate that the optimization approach leads to improved decision making capabilities by providing a suite of well planning solutions. This work highlights the advantage of utilizing the polynomial-based approach — the computational cost from multi-objective optimization can be substantially reduced in comparison to traditional approaches, which makes the proposed workflow more practical, rapid and repeatable, thus able to assist the decision making process in large field development.