Today, the role of large language models (LLMs) is undergoing a fundamental paradigm shift: from static text generators, they are becoming proactive, autonomous agents. These models are already capable of making decisions based on complex logic, interacting with software environments through purposeful tool-use, and interpreting the visual world through multimodal inputs. This development immediately raises the question of how this cognitive capacity can be applied to human movement in physical space. for controlling robotic devices? Modern robot simulation frameworks provide an ideal basis for validating the technology, in which the decision-making mechanisms and control stability of LLM-based agents can be analyzed through individually configurable scenarios, without physical constraints or costs.

The task is to design and implement an LLM agent capable of controlling the behavior of a physical agent in a simulation environment across different scenarios.

The student's details of the tasks:

• review of current solutions, literature research,
• examination of the simulation environment and the connectivity of LLM agents,
• design and implementation of the control agent operation,
• development of the LLM agent-based implementation of control in the selected simulation environment,
• Investigation of LLM agent behavior using various case studies.

The topic can be initiated within the framework of a permanent laboratory and continued as a thesis in subsequent semesters. The topic can also be carried out in a team.

Required competencies

To successfully solve the task, elementary programming knowledge (Python), basic robot control knowledge, and basic knowledge of programming (prompting) of large language models are required.

References for information

1. Antropic: Model Context Protocol (MCP)
2. OpenAI Agents SDK &és PydanticAI
3. Cyberbotics: Robotics simulation with Webots (https://cyberbotics.com/)
4. ALRM: Agentic LLM for Robotic Manipulation (https://arxiv.org/pdf/2601.19510)