This AI paper presents SRDF: a self-refined data flywheel for high-quality vision and language navigation datasets
Vision and language navigation (VLN) combines visual perception with natural language understanding to guide agents through 3D environments. The goal ...