Vision-IMU based collaborative control of a blind UAV

Position estimation of UAVs is usually done using onboard sensors such as GPS and camera. However, in certain practical situations, the measurements of both the GPS and the onboard camera of the UAV might not always be available or reliable. This paper investigates the possibility to overcome situations in which a UAV is blind, i.e. situations in which its onboard position sensors and estimates fail, by utilizing vision data from an external UAV. The paper demonstrates that by means of a collaborative approach, aimed at fusing information of multiple aerial vehicles, it is still possible to control the position of an UAV with onboard sensor fail. The method makes use of vision measurements from an external UAV in combination with the on-board IMU measurements of the main UAV to create a reliable position estimate. Indoor experiments demonstrate that position control is possible using this estimate.