Parameter estimation of social forces in pedestrian dynamics models via a probabilistic method

Area of Science:

• Physics
• Computational Science
• Social Science

Background:

• Crowd dynamics modeling is crucial for safety and urban planning.
• Accurate parameter estimation and model validation are key challenges in crowd dynamics research.

Purpose of the Study:

• To present a Bayesian probabilistic method for estimating crowd model parameters and their uncertainties from experimental data.
• To introduce a fitness measure for classifying and validating different crowd dynamics model structures.

Main Methods:

• Utilized real-life experiments and measurements for data collection.
• Applied Bayesian probabilistic methods for parameter estimation and uncertainty quantification.
• Developed a fitness measure for model comparison and selection.

Main Results:

• Successfully estimated model parameters and their probability density functions from experimental data.
• Demonstrated a method to assess the fitness of different crowd dynamics models.
• Laid groundwork for a general probabilistic model-selection strategy.

Conclusions:

• The proposed Bayesian approach provides robust parameter estimation and uncertainty quantification for crowd dynamics models.
• The developed fitness measure facilitates objective model selection and validation.
• This work advances the application of probabilistic data analysis in crowd dynamics research.