Abstract
This paper studies cooperative visual environmental monitoring based on gradient descent techniques on SO(3). We consider the scenario that vision sensors located in 3-D space control their orientations in order to optimally monitor 2-D environment. Then, the decision variables are constrained on the Lie group SO(3). Hence, we formulate the monitoring problem as an optimization problem on SO(3) and present its solution based on the gradient descent algorithm on SO(3). The proposed solution does not require density, an indicator of importance, defined on the environment but relies only on the density on the acquired image data. This allows one to apply the present method to changing unknown environment. In this paper, we also present a procedure to estimate the image density using current technology and examine real-time feasibility of the estimation process. Finally, we demonstrate the effectiveness of the proposed approach through simulation.
