VLAI: Exploration and Exploitation based on Visual-Language Aligned Information for Robotic Object Goal Navigation

Open Webpage

Haonan Luo, Yijie Zeng, Li Yang, Kexun Chen, Zhixuan Shen, Fengmao Lv

Published: 2024, Last Modified: 16 May 2025Image Vis. Comput. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Frontier object descriptions aid holistic environmental perception and enrich semantic agent knowledge.•Chat Describer helps the agent understand the spatial relationships for finding the target object.•The semantic similarity of frontier objects and the target object promotes exploration efficiency.