Affordances-Oriented Planning using Foundation Models for Continuous Vision-Language Navigation