Thesis

Perception and Prioritization of Visual Cues in Wood Ants

Organisms that rely on vision for navigation through their environment often do so by integrating many visual cues in order to guide themselves to their desired goal. In ants, research has identified a number of visual cues and features that guide an ant when navigating; however, how the objects are perceived or prioritized in a natural environment is unknown. One gap in the research is how ants, when there are several visual objects, extract different visual features and subsequently how these cues are behaviorally weighted. The ability to weigh the relevance against one another would allow the ants to prioritize a specific visual object or combination of the objects that can maximize their ability to reach a goal. In order to address this gap in knowledge, a model has been developed that explores how an ant can use different cues or combination of cues to facilitate navigation. To test our model, a combination of simulated and camera-imaged environments were examined during the initial random forging walk to locate a food source and a subsequent walk from nest to food site. Specifically, we have taken views from an ant’s visual perspective and processed these views using an image processing algorithm that extracts the geometric features, such as peaks, edges and centroids, that ants are known to use to guide their routes. As an ant progresses through the environment on its first random walk, features were extracted and only these features were stored in a “memory matrix” representing the sum total of features experienced during the walk. To test the value of the cues stored during the random walk, a second walk was performed from nest to food source. On the second walk ants would take a “snapshot” of their current view and extract the visual features. These features were then compared to the sum total of features stored in the “memory matrix” using pixel matching. The results from both the field studies and simulations suggest that in order to have the most robust success in foraging within the model, a global cue is necessary to allow the ant to get within the proximity of the goal and then they should switch to a local cue, which allows for more precise localization of the goal. Ultimately the work done here has allowed us to understand how visual information is prioritized and perceived as well as potentially expanding on how to increase the efficiency of cognitive processes such as memory storage and efficiency.

Relationships

Items