From Code to Motion: Building an Autonomous Hat-Hunting Robot with Kubernetes & ML

Get ready to bring a machine to life - literally. In this session, we’ll take you through the end-to-end journey of building an autonomous Raspberry Pi-based robot that can spot a red hat and drive straight toward it. Starting with a vision model trained on diverse images of hats, we’ll explore how to serve ML models efficiently using Kubernetes. Setting up Skupper then allows us to have secure service communication to the robot.

We’ll show how a Python application interfaces with the robot’s camera and the object recognition service to convert visual detection into real-world motion control.

Finally, we’ll demonstrate how to package both the app and ML services into an ARM image with Tekton Pipelines and deploy it directly to the robot via ArgoCD GitOps - enabling it to run fully autonomously without cloud dependencies.

Whether you’re a developer, ML engineer, or Kubernetes enthusiast, you’ll leave with practical insights on building intelligent edge devices that bridge the gap between software pipelines and physical automation.