Tracking Pedestrians across Multiple Microcells Based on Successive Bayesian Estimations

Abstract EN

We propose a method for trackingmultiple pedestrians using a binary sensor network.

In ourproposed method, sensor nodes are composed of pairs ofbinary sensors and placed at specific points, referred to asgates, where pedestrians temporarily change their movementcharacteristics, such as doors, stairs, and elevators,to detect pedestrian arrival and departure events.

Trackingpedestrians in each subregion divided by gates, referredto as microcells, is conducted by matching the pedestriangate arrival and gate departure events using a Bayesianestimation-based method.

To improve accuracy of pedestriantracking, estimated pedestrian velocity and its reliability in amicrocell are used for trajectory estimation in the succeedingmicrocell.

Through simulation experiments, we show that theaccuracy of pedestrian tracking using our proposed methodis improved by up to 35% compared to the conventionalmethod.