Assisted Agent-Based Simulations: Fusing non-player character movement with Space Syntax

Agent-based simulation is one of the core tools of spatial analysis utilised to provide an understanding of space when complex parameters come into play, such as how the visible space changes while traversing a building, or what happens when there is a destination to be reached. This type of simulation has a lot in common with techniques used in video games to create movement trajectories for non-player characters. Although these techniques have been developed over the years to provide more realistic and more “human-like” behaviour, they are rarely woven back into analytical and simulation tools. As a first step to remedy that, we developed a new methodology that fuses non-player character movement from computer games with simulation techniques traditionally used for agent-based analysis in Space Syntax. This first attempt utilises a different type of underlying representation of space, known as a navigation mesh. We first examine in detail two traditional techniques utilised in depthmapX agent-based analysis and highlight their strengths and limitations. We then describe how this technique differs from the classic space syntax methods, as well as how it can be combined to create hybrid analytical models of movement. The hybrid model developed in this case is that of a classic space syntax agent assisted by the aforementioned technique. We then tested and evaluated the traditional and new models for their capacity to explore two gallery spaces. The results extracted from the new hybrid simulation model depict agents with more capacity to explore, a significant addition to the traditional space syntax agent based methods