An Optimization-based Approach to Time Critical Cooperative Surveillance and Coverage with Unmanned Aerial Vehicles

The use of autonomous air vehicles equipped with cameras for surveillance is a natural but challenging application of robotic technology. Unlike ground robots, aerial platforms have many more operating constraints and their dynamic response typically dictates the nature of their role in such an application. While rotary-winged platforms enable more diverse maneuver options than their fixed-wing counterparts, maneuverability along a three-dimensional flight path imposes significant constraints. There is a class small, electrically powered fixed-wing aerial vehicle that feature body fixed cameras to provide a low-overhead platform with which to conduct surveillance operations. Their advantages are that they are portable, quiet, difficult to spot when aloft, inexpensive to acquire and operate, feature rugged airframes and are as safe as flight vehicles can be as their mass is less than 10kg. Their primary disadvantage in this application is that their flight path and sensor line of sight are highly coupled. We would like to develop a path planning capability to enable this class of vehicles to perform missions arguably outside of their design envelope. The benefits would be a low-cost, portable solution capable of being transported by ground personnel on foot. The system can, therefore, be considered to be mobile. The constraints imposed by selection of this type of air vehicle are summarized in ”Table I”. These vehicles have inner loop controllers that provide the requisite stability and control for the navigation autopilots they use so that waypoint navigation using GPS is reliable spatially but temporal performance is affected by ambient winds. Excessive maneuvering can degrade the precision of flight path control. Turns are always coordinated because of the constraint on sideslip. The coupling between flight path and camera for forward looking camera is illustrated in the ”Fig.1” where the motion of the camera frame at half-second intervals is shown as the flight path is varied from straight to turning flight over a 12 second period for the vehicle at the highest tier. The vehicle position, camera line-of-sight vector intersection with the ground-plane (center of image), and the camera frame edges on the ground-plane are shown. Aircraft roll dynamics impose a 0.5 sec time constant on roll rate commands and the vehicle turns in response to its