Semantic orientation for indoor navigation system using large language models.

Area of Science:

• Robotics
• Artificial Intelligence
• Computer Vision

Background:

• Autonomous robots are crucial for indoor navigation but lack seamless human interaction and environmental semantic understanding.
• Current systems face limitations in interpreting complex indoor spaces and collaborating with humans effectively.

Purpose of the Study:

• To develop an AI-driven object recognition system for autonomous indoor navigation.
• To enhance human-robot collaboration and semantic understanding of environments using Large Language Models (LLMs).

Main Methods:

• Integration of vision-based mapping with natural language processing (NLP) and interaction capabilities.
• Leveraging multimodal input and vector space analysis for enhanced perception.
• Utilizing advanced LLMs like GPT-4 Vision and Gemini for context-aware responses.

Main Results:

• Achieved enhanced object recognition and semantic embedding in indoor environments.
• Enabled more intuitive collaboration between humans and autonomous robots for navigation tasks.
• Demonstrated context-aware responses for improved spatial understanding.

Conclusions:

• The proposed AI-driven system, enhanced by LLMs, offers a novel framework for autonomous indoor navigation.
• This approach significantly improves spatial understanding and dynamic interaction in complex environments.
• Sets a new standard for seamless human-robot collaboration in navigation.