RxR-CE

The 'RxR-CE' dataset/benchmark contains a set of navigation tasks designed to evaluate the performance of agents in understanding and executing high-level semantic instructions within a spatial environment.

Papers using RxR-CE (3)

Path-level Hindsight Instructions for Semantic Exploration in Vision-Language Navigation2026One Agent to Guide Them All: Empowering MLLMs for Vision-and-Language Navigation via Explicit World Representation2026STRIDER: Navigation via Instruction-Aligned Structural Decision Space Optimization2025