Waseda University

“We use gpt-5.5-2026-04-23 and gemini-3.5-flash as VLMs, setting the temperature to 0.2 for both models and evaluating them independently.”

“The paper introduces a framework called Evidence Acquisition and Feasibility Gating (EAFG), which forces a robot to actively explore its environment to verify that required objects exist before it commits to a full task plan.”