Estimating Time-Dependent OD-Matrices for Pedestrian Infrastructures From High Frequent Pedestrian Counting Data

This paper proposes the estimation of origin-destination (OD) matrices depending on influence factors such as the time of the day from high frequent entry and exit counts at a pedestrian infrastructure. Estimation is based on explicit models for the temporal dependence, where the models are adapted from the dynamic freeway OD-matrix estimation approach. Since pedestrian counts are subject to non-negligible measurement errors, the estimation uses the generalized method of moments (GMM) estimation scheme to account for the errors-in-variables problem. A suitable estimation procedure is outlined. In a simulation exercise the method is shown to outperform recursive estimators, nonparametric approaches and Kalman filtering. Finally the method is applied to a case study in an Austrian shopping center. The evaluation of the accuracy of the method shows that the confidence bands are relatively large for the accuracy of the pedestrian counting sensors used. Advances in sensing technology will improve the accuracy of the counts in the near future and consequently increase the potential of the proposed approach.