Prioritizing Unaided Human Search in Military Simulations
Abstract
Search and Target Acquisition (STA) in military simulations is the process of first identifying targets in a particular setting, then determining the probability of detection. This study will focus on the search aspect in STA, particularly with unaided vision. Current algorithms in combat models use an antiquated windshield wiper search pattern when conducting search. The studies used to determine these patterns used aided vision, such as binoculars or night vision devices. Very little research has been conducted for unaided vision and particularly not in urban environments. This study will use a data set taken from an earlier study in Fort Benning, GA, which captured the fixation points of 27 participants in simulated urban environments. This study achieved strong results showing that search is driven by salient scene information and is not random, using a series of nonparametric tests. The proposed algorithm, using points of interest (POIs) for the salient scene information, showed promising results for predicting the initial direction of search from the empirical data. However, the best results were realized when breaking the field of regard (FOR) into a small number of fields of view (FOVs).
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2011
- Accession Number
- ADA547847
Entities
People
- James K. Starling
Organizations
- Naval Postgraduate School