3D Semantic Instance Maps: Leveraging Foundation Models for Language-Guided Navigation

Introduces O3D-SIM, an Open-set 3D Semantic Instance Map that uses CLIP/DINO embeddings for robust Vision-Language Navigation