Function-Based Rag for Vietnamese Speech Comprehension in Service Robotics

Service robotic speech comprehension is a fundamental capability in service robotics that plays a crucial role in applications involving interaction between humans and robots. While there have been significant improvements in speech recognition and natural language understanding technologies in recent years, there remains a notable gap in the application and assessment of these advancements for household service robots, especially concerning the Vietnamese language. Although the combination of Large Language Models and RetrievalAugmented Generation techniques has been extensively explored in chatbot technology, its application in service robotic speech comprehension, particularly within the Vietnamese language domain, has received little attention and has seen very few relevant studies. This paper discusses the implementation of a new method that applies the Retrieval-Augmented Generation technique with Large Language Models to assess speech comprehension in service robots, specifically tailored for the context of the Vietnamese language. The results are compared with the most widely used methods in the robot production market.