Using Evolving Curves to Track Dynamic Boundaries

The key innovation proposed in this STTR effort by Intelligent Automation, Inc. and Duke University is a hybrid level set based algorithm to achieve dynamic perimeter surveillance within a region by constructing an evolving function based on the perceiveddensity of a phenomenon. Drawing on state-of-the-art image processing methodologies, the hybrid solution offers the flexibilities of an Eulerian approach and the natural advantages of a Lagrangian one. In order to handle dynamic perimeter topologies, wecharacterize the boundary of the region to be monitored as a level set of the evolving function. To ensure validity of data within the region, the robots measure values of the quantity of interest and maintain and propagate an estimate of the evolvingfunction through a finite-element mesh. Every vertex in this mesh corresponds to a moving robot whose trajectories are described in a Lagrangian manner. A key aspect of the proposed innovation is a distributed agent-based implementation of the hybrid-levelset algorithm for tracking and propagation of the information across the mesh. We represent the communication between the robots using an ad hoc mobile network and describe a mechanism that handles data collection even when connections between specificrobots are broken. With the increasing threat of use of chemical and biological weapon, the need for technology to identify, mark out, and track regions of possible harm has increased. The proposed approach to perimeter surveillance provides a solutionto address this problem. The generality of the approach makes it viable for implementation on swarms of various types of robotic vehicles. Depending on the application the algorithms can be implemented on UGVs to track mine fields, surface and underwatervehicles to detect oil spills or even UAVs to aerially track chemical or oil spills, or to track large crowds or troop movement. Our initial customer for the proposed technology will be the DOD. We also anticipate marketing this technology to NASA for deepspace missions and planetary surface explorations where perimeter surveillance is critical.