| name | fleet-deployment |
| description | Deploy trained robot policies to edge fleets via FluxCD GitOps, image automation, and deployment gating |
Fleet Deployment
Deploy trained robot policies to edge robot fleets using FluxCD GitOps pipelines, automated image updates, and deployment gating.
Prerequisites
| Tool | Requirement |
|---|
kubectl | Authenticated to target cluster |
flux | FluxCD CLI 2.x |
az CLI | Azure authentication for ACR access |
Deployment Workflow
Step 1 — Bootstrap FluxCD
fleet-deployment/gitops/bootstrap.sh
Installs Flux components on the target cluster and configures Git source reconciliation.
Step 2 — Configure image automation
Define ImageRepository, ImagePolicy, and ImageUpdateAutomation resources in fleet-deployment/gitops/image-automation/.
Step 3 — Set up deployment gating
Configure gate criteria in fleet-deployment/gating/ to validate models before rollout.
Step 4 — Deploy to fleet
FluxCD reconciles cluster state from Git. New model images trigger automated manifest updates and gated rollout.
Key Directories
| Directory | Purpose |
|---|
fleet-deployment/gitops/ | FluxCD manifests, sources, releases, and cluster overlays |
fleet-deployment/gating/ | Deployment gating service and Kubernetes manifests |
fleet-deployment/inference/ | On-device inference runtime code |
fleet-deployment/examples/ | Example deployment configurations |
fleet-deployment/specifications/ | Domain specification documents |
Specifications