NavCoT: Boosting Vision-and-Language Navigation with Disentangled Reasoning
NavCoT introduces a novel strategy for Vision-and-Language Navigation (VLN) by enabling self-guided navigational decision-making through disentangled reasoning, leading to significant performance improvements.
