Each column shows one of 4 of the 10 cameras at four time instants denoted by k. A target marked with a box is tracked using a Kalman Consensus filter. The camera PTZ parameters are controlled so as to have the target continuously imaged at a high enough resolution.
In this project, we are developing methods for collaborative and opportunistic sensing in distributed camera networks. Consensus-based distributed estimation algorithms have been developed for target tracking and activity recognition. Camera network control has been realized using distributed optimization approaches related to potential games. The project involves active collaboration with Prof. J. A. Farrell of Electrical Engineering at UCR.
The project is currently funded through ONR grant N000140910666 (Mathematics, Computers, and Information Research). It has been supported in the past by NSF grant ECS-0622176 (Power, Controls and Adaptive Networks) and ARO grant W911NF-07-1-0485 (Computing and Information Sciences Division).

Demos

Cooperative control of a a camera network to focus on a particular target of interest (shown in red box)

Sample Publications