SurveillAnts

RESEARCH

 

A preliminary research has been conducted on different ant species, ant behaviors, caring for captive ants, and ant farms. In addition, various labs at NYU and Rockefeller University were consulting during the making of the project. Observing the ants also enabled us to learn more about their special behaviors, such as burying their dead, sensitivity to air movement, and deliberate groupings in specific spots outside of the nest.

 

Red Harvester Ants are between 5 and 7 millimeters long. They prefer arid, chaparral habitats and are native to the Southwestern United States (we got ours from Arizona). "Nests are made underground in exposed areas and can be up to 2.5 meters deep. Harvester Ant diets consist primarily of seeds, and they consequently participate in myrmecochory, an ant-plant interaction through which the ants gain nutrients and the plants bene t through seed dispersal״ (Wikipedia). We selected Red Harvester Ants for their size and for their quick tunneling skills.

HOW IT WORKS


Ant movements are being detected and recorded through a Logitech webcam that feeds directly into Processing. We are using a modified Open CV library for blob detection, and custom image processing code, in order to identify and track individual ants. Each ant, as it is detected, is provided with a unique id which is rendered in different colors inside the Processing sketch. Finally, this sketch is projected from above, back onto the ant's environment using the MadMapper projection mapping software.
The following picture shows what data the code is processing:

 

In addition to live rendering the ants' movements in real time, the program also records and stores their positions over time, capturing a history of past ant trails. To allow users to explore these trails and the patterns they create over time, we built custom knobs and buttons that allow people to control how much time is being revealed and how it is rendered onto the ant environment.

TECHNOLOGIES USED


Online data from a Webcam was processed using Blob detection library created by Julien Gachadoat and an image processing blur function by Mario Klingeman in Processing. Visuals created in Processing and were projected using Madmapper. Controlled box handled by Arduino.

Made with the kind support of: