The rapid development of artificial intelligence and robots, the research and development of intelligent dialogue robots are very necessary for today's society. However, a robotic algorithm system that simulates navigational English conversation has not yet been developed. In order to find a suitable algorithm system for dialogue robots, this paper uses the test data set to test the analytical model of navigational English dialogue instructions. The experimental results show that the conditional random field (CRF) + domain dictionary + ambiguity resolution method has the highest segmentation effect. The calculated percentages of the analytical model are correct rate: 76.85%; recall rate: 80.36%; F-value: 88.46%. This paper implements a robot teaching and reproduction method based on simulated navigational English conversation and human-computer interaction under the background of artificial intelligence, and designs robot motion realization experiments and speech recognition experiments. The three-dimensional error after fine-tuning the voice is between 1.6798mm and 2.9968mm. This article constructs a simulation navigational English dialogue robot system. The FAQ component has up to 79.2%; others have a lower accuracy rate of only 59.03%.