Autonomous landing algorithm using a sun position predicting model for extended use of solar powered UAVs

In the field of robotics, a major challenge is extending the flight range of micro aerial vehicles. One way to extend the range is by charging batteries with solar arrays on the ground, while resting on intermediate landing positions. The solution we propose in this study differentiates itself from other solutions as it does not focus on improving UAV efficiency but rather on finding the most efficient landing position. In particular, an algorithm is developed to show the usefulness of the approach. This algorithm makes uses of the sonar sensor on board of the Parrot Bebop 1 drone in combination with an OptiTrack system to scan the environment for potential landing opportunities. After these measurements are discretized on a 2D grid, analysis is carried out with a sun position predicting model. Finally, a landing position is chosen within the scanned area and the drone will land accordingly. Little is known on whether a solar powered charge on the ground could be effective in a limited period of time. We present a coarse analysis, showing that the DelftaCopter with solar arrays on its wings charges its batteries in 1.3 days with relatively cheap solar cells in Africa or Australia. Future work includes the use of computer vision instead of sonar as well as the ensurance of a safe landing position using vision.