Back to Community

v0.2.0 ยท Changelog

Version 0.2.0

DroneForge changelog notes for the v0.2.0 release.

DroneForge Nimbus App v0.2.0 is a bigger step toward a complete autonomous drone development workflow. It brings a UI overhaul, AI vision processing, cloud agent storage, hardware auto-discovery, live agent execution, validation, and better camera handling into Nimbus.

The result is a smoother path for building AI drones: create an agent, connect hardware, inspect the graph, run perception on the live camera feed, and deploy into a real drone loop with fewer manual steps.

Announcement

Nimbus v0.2.0 adds AI vision processing, cloud-backed agent storage, hardware auto-discovery, and a redesigned interface for building and deploying drone agents.

With these pieces together, Nimbus starts to feel less like separate tools and more like one connected Nimbus App for development drones.

AI vision processing

You can now enable YOLOv11 object detection or DepthAnything depth estimation on the live camera feed.

For developers building autonomous drones, that opens the door to more useful agent behavior. Instead of treating the video stream as a passive preview, Nimbus can process the feed and make perception available as part of the drone-development workflow.

Cloud agent storage

Agents can now be saved to and loaded from the DroneForge cloud.

That means your agents can carry progress across sessions instead of living only on one local machine. It is especially useful when you are iterating on AI drone behavior over multiple sessions, switching between machines, or moving from early testing into a more repeatable deployment workflow.

Hardware auto-discovery

v0.2.0 adds a simpler scanning and connection interface for cameras.

Instead of manually hunting through devices, Nimbus can scan available hardware and let you connect through a cleaner clicking interface. Multi-camera support also makes it possible to switch between connected cameras during flight operations.

Live agent execution

Agents can be loaded from the cloud carousel and deployed directly to the drone in real time.

The Deploy flow is faster because of it. A developer can move from saved agent to live execution without rebuilding the setup every time.

Agent validation

Nimbus now validates graphs before deployment.

Automatic checks catch common problems earlier, including type issues, invalid connections, and cycles. For drone programming, this matters. The safer place to find graph mistakes is before the program is sent into a live drone workflow.

Improved

  • Attitude HUD now delivers faster telemetry updates with sub-millisecond latency.
  • The Create tab, formerly Compose, syncs changes to the cloud automatically.
  • The Deploy interface shows available agents in a scrollable carousel for quick selection.
  • Multi-camera support lets developers switch between connected cameras during flight operations.

Release context

This changelog entry is part of the DroneForge release archive for Nimbus and DF1. The archive helps builders understand how the autonomy stack has changed over time, including updates to setup, Nimbus App behavior, Python Library workflows, documentation, hardware notes, flight workflows, and community-facing product polish.

Older releases are still useful when comparing versions, debugging a local installation, or following the path from early drone connectivity work to current autonomous flight tooling. Use these notes as historical context alongside the current documentation, Nimbus App releases, and Python Library updates.

For current projects, compare these archived notes with the latest release before making setup decisions. DroneForge continues to refine how developers connect hardware, stream telemetry, inspect video, run agents, and build repeatable autonomy workflows.

Community archive

Continue exploring DroneForge changelogs, research notes, and Nimbus examples through the community archive. These internal links help connect related releases, technical notes, and builder resources.