SIMT

SIMT ECHO is an offline-first spatial navigation system that transforms a short smartphone walkthrough into a portable, shareable spatial memory for others. Built on a Kotlin Multiplatform architecture for Android and Wear OS, the platform allows one person to record a path through an unmapped or GPS-denied physical space and transfer that route to another user, who can navigate the exact path completely offline in Airplane Mode without pre-existing floor plans or external beacons. The platform divides navigation intelligence into three distinct layers: • Deterministic Sensor Truth: Physical movement, step counts, turn geometry, and relative distance are strictly captured and calculated by on-device Inertial Measurement Unit (IMU) sensors and Pedestrian Dead Reckoning. • Meta AI Visual Grounding: With creator consent, keyframes are processed by cloud vision models—including Llama 4 Scout, Muse Spark 1.3, and Segment Anything Model (SAM 3.1)—to detect, segment, and label permanent visual landmarks without fabricating spatial coordinates. • Probabilistic Local Guidance: During replay, an along-route Ghost particle filter and the ECHO Parallax monocular visual engine fuse real-time step events, turn signatures, level changes, and visual view-strips to track progress and display an honest uncertainty aura.