Virtual agent based static and dynamic boundary estimation and tracking with multiple agents

Abstract

In this paper we have proposed a virtual agent based cooperative boundary tracking and estimation strategy. It is assumed that agents are situated in a fixed interaction topology, where each agent can communicate with its fixed forward and backward agents. The proposed boundary tracking algorithm is derived by optimizing the majorization function of the weighted stress function. It is shown that the algorithm is globally asymptotically stable. Using this algorithm agents cooperatively track the boundary. The whole task is divided into two phases. In the first phase, except the first agent, all are following it's previous actual agent with the help of it's virtual agent. In the second phase agents converge to the boundary and start estimating the boundary. Each estimated boundary point is equidistant from the two boundary points measured by the two fixed neighbours. Virtual agents are used to estimate the boundary points. In this phase they surround the boundary and track it in the counter-clockwise direction. To our best knowledge, the virtual agent concept has been used for boundary estimation and tracking for the first time. We simulate our algorithm for static and dynamic boundary.