Vision-Aided, Cooperative Navigation for Multiple Unmanned Vehicles

The motivation of this research is to exploit three attributes of increased unmanned vehicle use for intelligence, surveillance, and reconnaissance missions. These attributes are: increased numbers of unmanned vehicles, on-board vision, and wireless communications. The research begins with the development of a cooperative navigation system based on the measurement of vehicle position relative to shared landmark position estimates. Each vehicle in the network locates landmarks using it's on-board vision system and transmits the data to all other system vehicles. After receiving data from the other vehicles, the system fuses the landmarks with on-board measurements using a federated filter architecture. Simulations of the cooperative system, with and without ranging, are compared to a non-cooperative simulation. The comparison is performed using four platform motion scenarios: stationary, linear, angular, and full motion. The simulation results demonstrate position error estimate improvements of 0.5cm to 1cm. Additionally; the stationary and linear motion scenarios demonstrate attitude observability difficulties eliminated by the introduction of angular motion.