Flight path planning for unmanned aerial vehicles with landmark-based visual navigation

• The flight performance and the navigational capabilities are considered.
• The airspace is discretized by a network depending on the UAV characteristics.
• The time-consuming tasks are performed in a preprocessing step.
• It is demonstrated that a flight path is calculated within few seconds.

In this paper we present an algorithm to determine a shortest trajectory of a fixed-wing UAV in scenarios with no-fly areas. The innovative feature is that not only the kinematic and dynamic properties, but also the navigational capabilities of the air vehicle are taken into account. We consider a UAV with landmark-based visual navigation, a technique which can cope with long-term GPS outages. A navigation update is obtained by matching onboard images of selected landmarks with internally stored geo-referenced images. To achieve regular updates, a set of landmarks must be identified which are passed by the air vehicle in a proper sequence and with appropriate overflight directions.The algorithm is based on a discretization of the airspace by a specific network. Each path in the network corresponds to a trajectory which avoids the no-fly areas and respects the flight performance of the air vehicle. Full functionality of the navigation can be ensured by dynamically adapting the network to the environmental conditions. A shortest trajectory is then obtained by the application of standard network algorithms.