Agentic Workflow for Robotics

I presented my embodied AI agent project, which combines an LLM, a vision language model, and a VLA model to build semantic maps and let a robot navigate on its own. I walked through how I connected these foundation models to ROS2, how robot actions get tokenized, and how groups like Google DeepMind and Nvidia are approaching embodied intelligence.