Multi-Window Cockpits in Android Automotive: From CTS Battles to Declarative UI

Building a multi-window infotainment cockpit in Android Automotive used to be hard. To place several apps side by side and animate them, you had to go deep below the application layer: custom TaskDisplayArea orchestration, SurfaceControl animations frame by frame, unpredictable z-ordering, and a CTS run that broke the moment you touched window layering.

Scalable UI changes that. A full multi-window cockpit becomes a composition of orchestrated panels, declared in XML through RRO overlays. No glue code, and the orchestration that once required deep internals is handled by the framework. What took months of fighting the Window Manager now takes hours.

This is not a walkthrough of the docs. It is the view of an engineer who built cockpits the hard way before Scalable UI existed: what changed, the trade-offs the docs don't mention, and how far you can push the framework, from panels, variants and transitions to the point where XML ends and code begins. With live demos throughout.