Abstract: Development of context-aware technologies and rapid advancement in communication are essential parts of most industries, such as underground mining, pervasive medical care, smart space, and wireless sensor network surveillance. On occasions that require positioning services, location techniques offer convenience and may even save lives. In underground mining, not only do the miners work in a hostile environment but the environment itself threatens their lives. Thus, determining the precise location of people and objects in underground environments is essential. However, GPS cannot be used for practical application underground and underground localization seems to be the most feasible way to provide positional information to mining vehicles. The localization of underground vehicles has been a critical obstacle in the development of intelligent mining vehicles. Faced with the bad environment, interference, and multipaths among other effects, it is difficult to obtain high-precision positioning results using conventional algorithms. In addition, the underground environment is mostly made up of long narrow tunnels, which is not conducive to an arrangement of anchor nodes, and the layout of underground anchor nodes usually has great influence on the positioning results. Therefore, ordinary positioning methods do not meet the high-precision positioning requirements of intelligent mining. In this paper, traditional trilateration was analyze, the reasons for error was summarized, and an improved algorithm was proposed. Simulation results showed the effectiveness of the improved algorithm. In addition, the principle of topology optimization was proposed by theoretically analyzing the error band and using the maximum absolute positioning error to simulate the influence of topology on positioning accuracy. According to the characteristics of the environment, minimizing the average maximum absolute positioning error was the principle of topology optimization. Here, simulation and field experiments were carried out to verify the improved algorithm and topology optimization method. The experimental results show that the improved algorithm can reduce the error by 15% -43% under the same topology, the optimized topology can reduce the error by 17% -65%, and a combination of the two can reduce the error by 74%. The results show that under the same localization conditions, the proposed algorithm can significantly improve the accuracy of the localization results. In addition, there is a close relationship between localization result and topology structure. Based on the actual environment, choosing a flexible topology layout can further improve positioning accuracy, and combining the improved algorithm with the topology optimization method can achieve a higher positioning accuracy.