Lumina — Embodied Spatial AI
A production-grade Multi-Agent System (MAS) that gives visually impaired users real-time spatial awareness, object memory, and voice-guided navigation — powered by computer vision, monocular depth estimation, and LLM-driven agent negotiation.