300 members of a colony of Aphaenogaster cockerelli
Automated Tracking and Modeling of Social Insect Behavior
At the CMU
we are investigating algorithms for automatically
tracking and modeling the
behavior of multiagent systems.
We have selected social insects as research subjects because they
can be observed in a laboratory environment and
they provide a rich source of data for testing our algorithms.
Our goals are
Tracking the movement of hundreds of ants simulataneously.
This is a view of our experimental arena taken from a camera above the foraging ants. On the left is a raw image from the video stream. The center image shows moving ants highlighed with green bounding boxes. On the right we show the output of our analysis program that associates individual ant movement traces with the sequences of bounding boxes.
The system can monitor and log the positions and behavior of hundreds of individual ants at a time. In addition, the application of automated statistical tools to this data will enable us to model insect behavior more accurately and precisely than was previously possible.
A comparison of spatial activity in an arena with and without the presence of food.
In the MultiRobot Lab we are particularly interested in learning about social insects, as they provide an existence proof of successful large-scale robust behavior forged from the interaction of many, simple agents.
Ant behavior can offer a wealth of ideas on how to organize a cooperating colony of agents. As an example, even though they are only capable of very short range communication, ants are able to carry out complex scouting and retrieval operations over tens of meters. The techniques social insects utilize for staging such complex operations could also be employed in the design of robust multi-robot systems --- it is important for us to learn what insects have to offer.
Biologists are also beginning to draw on computer science. In a recent Nature article, for instance, Deborah Gordon suggested that the growth of theory in social insect research has been inspired by the artificial intelligence community. Accordingly, we believe the research in our lab can contribute substantially to the study of insect behavior; in particular, we have developed
We will extend these existing tools for the observation of insects in our laboratory. We will also develop new tools for modeling groups at a societal level rather than just the individuals in a society.
The animals being observed move about an arena monitored by an overhead color camera. The output of the camera is processed using color classification and region segmentation software developed in our lab (CMVision). The image processing step provides the locations of the animals being tracked.