Legged Open-Vocabulary Object Navigator (LOVON) is a novel framework jointly proposed by the Hong Kong University of Science and Technology Guangzhou and the Beijing Humanoid Innovation Center on July 10, 2025. It aims to use large language models for hierarchical task planning and combine them with open vocabulary visual detection models to achieve long-distance object navigation in dynamic and unstructured environments.LOVON: Legged Open-Vocabulary Object Navigator".

LOVON is designed for efficient long-distance navigation in dynamic, unstructured environments. It is particularly well-suited for scenarios requiring high flexibility and adaptability, and excels in legged robotics applications and long-sequence task execution.