Modular, low-cost arboreal ant (Azteca alfari) tracking sensor development in Panama

Monday, November 17, 2014: 10:24 AM
D137-138 (Oregon Convention Center)
Andrew Quitmeyer , Georgia Tech, Atlanta, GA
The intricate habitats of tree-dwelling ants pose difficult challenges for scientifically monitoring behavior. Conventional laboratory methods, such as video recording, are often appropriated for automated tracking, but the complexity of dynamic forest environments causes poor results in most post-processing computer vision software. Manually tracking these multitudinous creatures then slows and hinders scientific research. Instead, devices are needed that are custom built with scientists to grapple with the particular difficulties of tracking arboreal ants.

First, the sensors should interfere with the creatures as minimally as possible. The sensor also needs to be able to track the ants on arbitrary surface shapes due to the geometric complexity of the forest. Unlike many computer vision systems, this device must also deal with similarities in coloration between the ants and the trunks and branches upon which they walk as well as low light or changing light conditions.  The devices should handle environmental hardships like humidity and dirt and be inexpensive enough to replace when damaged. It needs to also be readily documented and hack-able for scientists to be able to adapt and customize the device to the exact needs of their experimentation. Finally, in order to be truly valuable as a scientific tool, the sensor should provide real-time feedback in the field to the scientist concerning its proper functioning.

Promising prototypes were already developed in previous field sessions which used differences in reflected infrared light. More iterations and evaluations during upcoming field seasons will hone the sensor’s design into a powerful research tool.