LifelongCrossNav for Multi-Floor 3D Semantic Navigation
August 10, 2026
LifelongCrossNav implements a shared sparse 3D semantic voxel memory to enable persistent multi-object navigation in unknown multi-floor environments. The framework uses support-aware traversability mapping and stair-specific perception to allow agents to retrieve scene information across different levels without rebuilding maps.
HOW THIS AFFECTS YOU
●
researcherYou can address the gap between multi-object and cross-floor navigation using integrated semantic voxel memory.