State-dependent and distributed pedestrian tracking using the (C)PHD filter

The use of the Probability Hypothesis Density (PHD) filter family for distributed indoor pedestrian tracking with laser scanners is discussed. A Sequential Monte Carlo (SMC) implementation with labeled particles is presented which avoids the need for particle clustering. A special focus of the proposed method lies on a state-dependent modeling of the sensor characteristics. The measurement-based proposed model incorporates changes in the probability of detection due to distance, occlusions and the sensor location dependent environment leading to superior tracking results in simulation and real experiments.