Abstract: This work presents a novel approach to zero-shot visual object goal navigation that leverages the ability of visual Large Language Model (vLLM) for finding target in unknown environment. Our ...