Speech comprehension in service robotics serves as a cornerstone functionality for various applications centered around human-robot interaction. Despite advancements in techniques and services for speech recognition and natural language comprehension in recent years, dedicated implementation and validation of these techniques on household service robots, particularly in the Vietnamese language domain, have been lacking. This paper elucidates the execution and outcomes of a practical benchmark, named the Vietnamese Robotic Speech Command Dataset (VRSCD). The VRSCD is designed for evaluating speech comprehension in service robotics. It also allows for the comparison of the performance of many popular large language models (LLMs), tailored specifically for the Vietnamese language context, to better serve the Vietnamese robotics industry.