ObjectNav

'ObjectNav' is a benchmark that evaluates an agent's ability to navigate and manipulate objects in unseen environments using language and visual observations to inform spatial actions.

Papers using ObjectNav (4)

Uni-LaViRA: Language-Vision-Robot Actions Translation for Unified Embodied Navigation2026Semantic Evidence Regulation via Relational Bias for Zero-Shot Object Navigation2026FiLM-Nav: Efficient and Generalizable Navigation via VLM Fine-tuning2025DyNaVLM: Zero-Shot Vision-Language Navigation System with Dynamic Viewpoints and Self-Refining Graph Memory2025