Description
The Floor Field model is a discrete-space crowd model — usually layered on top of cellular-automata — in which cells carry two scalar fields. The static floor field encodes distance to exits / attractors and biases agent transitions. The dynamic floor field is laid down by passing pedestrians and decays / diffuses, producing a stigmergic herding bias. Variants (wall floor field, interplay floor field) extend this with obstacle avoidance and pairwise interaction terms. The note folds all variants together because they share the same cell-transition mechanics.
When it's used
- Discrete-grid evacuation simulation
- Stigmergic herding behaviour without explicit communication
- Coupling with
cellular-automatato get computationally cheap large-crowd models - Generating density / flow fields for CSI cross-validation
Limitations
- Discrete grid size constrains spatial resolution
- Parameters (k_S, k_D, decay α, diffusion δ) need empirical tuning
- Less faithful to fine-grained body dynamics than continuous SFM
Source Papers
- porzycki2023_6cf3 ↗ — floor-field crowd evacuation study
- maury2018_d24a ↗ — review covering floor-field variants
- lu2026_fb07 ↗ — recent floor-field cellular-automaton model
- echeverrahuarte2023_fcc4 ↗ — floor-field-based density estimation
- kleinmeier2019_e6cd ↗ — Vadere-style floor-field implementation