High-fidelity behaviours for model people on model streetscapes

Abstract

Computational techniques for generating plausible-looking and authentically behaving synthetic walkers on simulated sidewalks are an important component of social science, engineering, planning, policy and management scenarios for many downtown settings, particularly for phenomena that are difficult to experiment with in reality. Yet, traditional approaches used to fashion model streetscapes and model people to fill them are usually sourced in proxy behaviours borrowed from physics and information heuristics, and while these are often conveniently parsimonious and analytically tractable, they can be limited in their ability to generate the sorts of rich synthetic realities in simulation that are commonly desired for extended experimentation. To facilitate the generation of higher-fidelity behavioural models, I present an alternative scheme based on behavioural geography in theoretical foundation and on polyspatial automata, space–time calculation and efficient computing as actionable media for simulation. Ultimately, the aim in introducing new approaches is to broaden the range of experimental scenarios that models can be applied to, and I present varied applied tests for the model infrastructure as a demonstration of the concept.