Safe and efficient mobile robot navigation remains a critical challenge in industrial environments, particularly in the presence of humans, whose movements are inherently unpredictable. To address this, we propose an AI-driven human-robot interaction (HRI) framework that integrates vision-language models (VLMs) and large language models (LLMs) to enhance navigation safety. Implemented in NVIDIA Isaac Sim, our system detects humans, enforces a minimum safe distance of 1 m, signals for attention, and awaits instructions before proceeding. If instructed to continue, the robot resumes movement at a reduced speed of 0.6 m/s. Experimental results demonstrate 85.7% human detection accuracy, 93% gesture recognition accuracy, and a recorded minimum distance of 0.93 m from humans. By enabling natural communication through vision and language, the system enhances robot adaptability and accessibility, facilitating seamless human interaction without requiring technical expertise. These findings underscore the potential of AI-driven navigation for safer and more collaborative industrial environments. The code used in this study is available open-source at GitHub Repository ....