A methodology to calibrate pedestrian walker models using multiple objectives

A methodology to calibrate pedestrian walker models using multiple objectives

Campanella, M.C.
Daamen, W.
Hoogendoorn, S.P.

Civil Engineering and Geosciences

Transport & Planning

2012-10-30

The application of walker models to simulate real situations require accuracy in several traffic situations. One strategy to obtain a generic model is to calibrate the parameters in several situations using multiple-objective functions in the optimization process. In this paper, we propose a general methodology for calibration of walker models. This methodology is a generalisation of existing calibration procedures adapted to walker models. The fundamental aspect of the methodology is the use of several scenarios representing different calibration objectives. One of the advantages of the general methodology is that by applying it, the process of calibration helps understanding themodel and how to adjust it according to the intended application. As an example, the methodology is applied with synthetic generated trajectory data using the Nomad model to investigate the influence of the mathematical specifications of the objective functions and the flow configurations in the accuracy of estimations and significance of individual parameters.

pedestrian models
calibration
multi-objectives

http://resolver.tudelft.nl/uuid:5a134471-0bda-4d39-be58-76def10f117e

TRAIL Research School

TRAIL Beta-Congress: Mobility and logistics - Science Meets Practice, Rotterdam, The Netherlands, 30-31 October 2012

Institutional Repository

conference paper

(c) 2012 The Author(s)
TRAIL Research School