Natural Language Navigation Task Allocation for Robot Based on Agents

In recent years, technological developments have driven remarkable achievements in the field of natural language processing (NLP), especially the rise of large language models (LLMs), which have had an important impact on the field of robotics. Before the development of LLM technology, robotic systems had significant difficulties in parsing the implicit intentions in natural language instructions, and it was difficult to effectively integrate environmental information with the robot’s state to achieve efficient planning and allocation of tasks. In the field of task allocation research, existing literature mostly focuses on executing task instructions with clear goals, while ignoring the importance of human-computer interaction. LLM-driven AI agent research brings task planning capabilities and natural language interaction capabilities with human users to physical robots. However, research on the AI agents and robots is still in its early stages. This study proposed an LLM-based Agent method that can integrate environmental information and robot status to reasonably plan and allocate natural language tasks and verify it in a real environment.