Estimation of Pedestrian Distribution in Indoor Environments using Multiple Pedestrian Tracking

[email protected] [email protected] Abstract We propose a two-tier data analysis approach for estimating distribution of pedestrian locations in an indoor space using multiple pedestrian detection and tracking. Multiple pedestrian detection uses laser measurement for sensing pedestrians in a heavily occluded environment which is usually the case with most indoor environments. . We adapt a particle filter based multiple pedestrian tracker to address the constraints of a limited number of sensors, heavy occlusion and real-time execution. Under these conditions any detection and tracking technique is likely to encounter a degree of error in cardinality and position of pedestrians. A completely new approach is employed which measures the error in tracker output due to occlusion and uses it to estimate a probability density function which represents the probable number of pedestrians located at a particular exhibit at a particular time. The end result of the system is a variable representing cardinality of pedestrians at a particular exhibit. This variable follows a distribution which is approximately normal where the variance of the probability distribution function is directly proportional to the error encountered by the tracker because of occlusion. The accuracy of our detection and tracking algorithm was tested both separately and in conjunction with the second-tier pedestrian distribution analysis and found marked improvement making our average pedestrian counting accuracy to at least 90% for all the pedestrian position data that we gathered with average pedestrian density at 0.34 pedestrians per sq. meter. Since the environment constraints for our system are unprecedented, we were unable to compare our result to any previous experiments. We recorded the number of people at each exhibit manually to establish the ground truth and compare our results.