A novel vision-language model-based approach for socially compliant robot navigation in human-centric environments.