Abstract: The application of vision-language models (VLMs) has achieved impressive success in various robotics tasks. However, there are few explorations for foundation models used in quadruped robot ...