Our lab studies collective animal behavior and granular materials. While these topics may seem very different, each are collections of many particles. The ‘particles’ in our lab are athermal, don’t conserve energy, which means that these systems are out-of-equilibrium. Its not so straightforward to connect how individual particles behave back to whats going on in the system as a whole.
My (Sam) research focuses on granular materials, and how can physics help us understand and ultimately predict earthquakes and avalanches. Some everyday examples of these materials are sand and salt. What makes granular materials so complex is how they can transition from a solid to a liquid (think salt pouring from the shaker). My research consists of analyzing how granular materials act under shear strain and building an apparatus to conduct this analysis in a controlled environment. I will be focusing on what happens at the particle scale and relating that back to behavior of the system. The granular materials in this experiment were PSM-4 (photo-stress materials), which are made specifically so that all contact forces made on each particle can be seen under polarized light. Linear actuators apply shear to a collection of approximately 1,000 particles. The particles are illuminated by a large LED light panel from below. An overhead camera snaps images of the particles. A stepper motor is mounted beneath the camera, which moves a polarizing filter in front of and away from the lens of the camera, so that we could obtain pictures of the particles with and without the force chains.
I’ve written code to automate the motion of linear actuators and a stepper motor, and remotely trigger and download images from a camera. Now, I’m learning to analyze the images, and track particles using hough transformations and other cool algorithms.
Collective Animal Behavior
I’m (Julia) researching collective animal behavior. Many different groups of organisms exhibit collective behavior, from the cells in your body, flocks of birds, to people living in cities, we can ask many of the same questions about their organization and behavior. How and why do motion patterns in collective systems arise? How do individuals in a group communicate? Can we mathematically model systems based on the individual parts or the global properties? Would such models be applicable to a wide range of systems (or species)? Could we then use those models to predict the behavior of groups of animals? In order to start to investigate some of these questions, we are studying cohesion in swarms of young Artemia. Using high-speed cameras and stereo-imaging techniques, we can obtain quantitative spacial data for each individual in a swarm.
After making necessary adjustments to the apparatus that ensure good video quality and high contrast between the brine shrimp and the background, we have been able to start processing videos using (primarily) computer code written in Matlab. First, the cameras must be calibrated in order to obtain accurate 3D data from the videos with different 2D perspectives. To do this, images of the following calibration mask are analyzed.
We chose brine shrimp as a test organism because they are relatively inexpensive to raise and they form swarms in response to light. In particular, brine shrimp have a strong phototactic response to blue-green wavelengths of visible light, so we are using green LED light sources to manipulate their swarming behavior. As mentioned, data collection involves utilizing high-speed cameras and stereo-imaging techniques. Several infrared lights are being used to illuminate the tank and the cameras because the Artemia cannot see or are impartial to longer wavelengths of light. The apparatus is kept in the dark to ensure that the brine shrimp are only responding to the stimulus of the green LED. In order to capture 3D data, we have constructed a hexagonal tank that holds the shrimp. Three cameras are positioned near adjacent faces of the hexagon; these different perspectives provide the necessary information to resolve a single 3D perspective of the system.
Every frame in each video from the three cameras is then altered so individual animals can be detected and their 3D position in the swarm can be resolved as time progresses. Information can be taken from each image based off of the contrast between a light brine shrimp and the dark background.
After each individual is detected, the stereo-matching is done which combines the 2D data from each camera into one 3D data structure. Until relatively recently, this kind of data was very hard to take. Each of the three cameras in the apparatus captures ten frames every second and each frame contains the position of hundreds of Artemia. Processing all of the information that we are collecting requires a lot of computing power. I am now stereo-matching the particles and tracking the Artemia in 3D. Soon we will take and analyze different sets of data that may provide insight into the dynamics of the collective behavior of brine shrimp.