Particle Filter for 1D Underground Localization Based on Pedestrian Dead Reckoning

An algorithm for tracking a pedestrian in an underground local positioning system is discussed. The mine structure is described as a graph. The inertial measurement unit restores the trajectory in real time. The trajectory is fed to the particle filter containing turn detector and edge detector modules. The algorithm effectively works with a small number of particles and can be implemented on a microcontroller. It is shown that such a small number of particles, like 10, is enough for the average positioning accuracy of approximately 10 meters for common structures of mines and tunnels with characteristic segment length of approximately 70-100 m.