This peer-reviewed article (Cognitive Computation, 2021) proposes an information-theoretic agent–signage interaction model to quantify and simulate how signage reduces uncertainty during human wayfinding. The authors ran two large online crowd-sourcing experiments (n=307 for visibility; n=171 for decision-confidence) to empirically derive visibility and decision-confidence distributions as functions of distance/angle, then refined and validated the model with a VR experiment (40 participants; training set 30 trajectories, test set 10 trajectories). The refined model reduced DTW distance between simulated and human trajectories by 38.76% and supports parameterizing agents by lookahead distance (LD=28) and threshold entropy (TE sampled from observed distribution).
Questions this document answers
10.1007/s12559-019-09689-1
https://doi.org/10.3929/ethz-b-000387187
Published online 17 December 2019; journal year 2021 (Vol.13:189–206).
307 participants (after exclusions: analyzed responses: 120 + 134).
171 participants (20 excluded).