Attention:

We are currently looking for social and behavioral science research assistants. Those interested can contact Nikos Papanikolopoulos

Researchers from the Artificial Intelligence, Vision, and Robotics Lab (AIVRL) have set up a system of mutliple Kinect RGB+D sensors in a research preschool classroom in the Shirley G Moore Lab school at the University of Minnesota. This system can be used to monitor the children in the classroom. Currently, this system consists of 5 Kinects that are connected to 3 PCs. The 5 cameras are strategically placed along the walls of the classroom so that each camera captures a novel view of the room. The cameras are then configured to begin recording at approximately the same time.

Each camera is calibrated so that it's exact position in the room is known as well as the characteristics of the camera itself. Using this calibration along with recorded RGB+D information from the Kinect a 3D representation of the room can be recreated. This 3D representation is then used to extract information about the objects in the scene. These objects are then identified and tracked as they move about the scene using region covariance descriptors.