Approach of personnel location in roadway environment based on multi-sensor fusion and activity classification

Abstract For the demands of localization system in underground mines, accuracy and generalized ability are the key indicators to evaluate the performance of the location algorithm, which directly affect the reliability and perception performance of the safety monitoring system. Since the irregular tunnel environment is dusty, humid and noisy, the reliability and generalized ability of the localization wireless network is an urgent and challenging problem. In this paper, multiple kinds of sensors are used to collect the data and the fusion mechanism is designed to improve the robustness of the location method. Considering the prior knowledge of the environment, a reliable activity classification method based on the Random Forest (RF) along with wireless Round-trip Time of Flight (RToF) technology to calibrate the accumulative error is proposed, which confirms the special position by matching with the known map of the underground. The positioning strategy with multiple sensors fusion enables the localization system adapting to kinds of harsh tunnel environment. The activity recognition performance is evaluated and numerical results indicate that the accuracy can reach more than 97%. Additionally, according to the simulation and experimental results, the impact of outside interference on the proposed method is substantially mitigated. The results demonstrate that the proposed method for localization system is practically feasible.