Designed for newcomers to the field, this session covers the core concepts of agentic tools and agent–robot boundaries. Participants will learn to safely control simulated robots, observe their behavior, and build simple agents that make decisions based on world state using foundational agentic frameworks.

No prior experience with agentic frameworks is required—only basic familiarity with robotics or AI concepts is needed. The session will cover:

• From LLMs to chat models to agents

• ReAct agents

• Common agent design patterns

  • Tools

  • Planning

The tutorial is accompanied by hands-on script files that allow participants to interact with ROS (turtlesim), giving them a lightweight, low-risk environment to experiment with agent-driven robot control.

Resources and Tools

Github Link - Session 1