Investigating player perception of environment-aware embodied conversational agents enabled by vision language models