Flyphone: Visual Self-Localisation Using a Mobile Phone as Onboard Image Processor on a Quadrocopter

An unmanned aerial vehicle (UAV) needs to orient itself in its operating environment to fly autonomously. Localisation methods based on visual data are independent of erroneous GPS measurements or imprecise inertial sensors. In our approach, a quadrocopter first establishes an image database of the environment. Afterwards, the quadrocopter is able to locate itself by comparing a current image taken of the environment with earlier images in the database. Therefore, characteristic image features are extracted which can be compared efficiently. We analyse three feature extraction methods and five feature similarity measures. The evaluation is based on two datasets recorded under real conditions. The computations are performed on a Nokia N95 mobile phone, which is mounted on the quadrocopter. This lightweight, yet powerful device offers an integrated camera and serves as central processing unit. The mobile phone proved to be a good choice for visual localisation on a quadrocopter.