Dependable automation
A command is only the beginning. What actually happened?
What I am examining
I investigate local voice, device coordination, and automation in homes and working environments. A device can acknowledge a command while its physical state remains unchanged. A sensor can report something that was true a minute ago. I study how the system recognizes and handles those differences.
Why it matters
People need to understand what an automated system did, recover when it fails, and take control when judgment is required. Those capabilities belong in the original design.
What the evidence must establish
I examine lost connections, stale readings, conflicting instructions, and incomplete execution. The useful measurements include unintended actions, recovery time, and whether a person can establish the actual state and regain control.
The comparison
Existing platforms already provide capable voice interfaces and device control. My research concerns the coordination around them: how observations become state, how uncertainty affects action, and how one failure reaches another component.