IoT-Brain: LLM Grounding for Proactive Sensor Scheduling Closes Semantic-to-Physical Gap

arXiv·low signalXBlueskyLinkedInCopy link

Formalizes Semantic-Spatial Sensor Scheduling (S3) — the proactive decision of what to sense and when — showing that direct LLM planning fails at this task. IoT-Brain introduces a grounding architecture that bridges the semantic-to-physical mapping gap, enabling intent-driven sensor operation instead of predefined monitoring schedules. Demonstrates that existing perception-centric IoT pipelines operate retrospectively and miss the fundamental planning decision of proactive sensing.