The DRS-GUI framework enhances GUI grounding for Multimodal Large Language Models by introducing a training-free approach that mimics human perceptual actions. This innovation results in a 14% performance improvement in identifying relevant UI elements, indicating a significant advancement in the field of AI-driven user interface interaction.