ORIGIN-DESTINATION MATRIX ESTIMATION FOR PEDESTRIAN NETWORKS WITH BI-DIRECTIONAL FLOW EFFECTS

This paper deals with the problem of origin-destination (O-D) matrix estimation from pedestrian counts. This problem is formulated as a bi-level programming problem. The lower-level problem is the user equilibrium (UE) assignment problem for pedestrian networks with bi-directional flow effects, while the upper-level problem is to estimate the O-D demands from pedestrian counts. To accommodate the effects of bi-directional pedestrian flows at various flow conditions, the walking time function with generalized bi-directional pedestrian-flow ratio (GBPR) is calibrated using data observed at an outdoor walkway in Hong Kong. The lower-level problem incorporates this GBPR function to take account of the bi-directional flow effects and is able to provide more realistic estimation of pedestrian network flows. Two solution algorithms are adapted for solving the proposed bi-level programming problem. A numerical example is provided to illustrate the applications of the proposed model and of the two developed solution algorithms. Algorithm 2 (diagonalization method with Newton method) can give solution with the lowest upper-level objective value than the Algorithm 1 (diagonalization method with method of successive average). The O-D estimation error will increase if the bi-directional flow effects were not taken into account in the O-D estimation process.