Visual landing of helicopter by divide and conquer rule

摘要

An imperative problem in the field of vision-based aerial robotics is the ‘autonomous helicopter landing’. The complexity of this problem can be eliminated by dividing the problem into sets of independent and simple sub-problems and to solve each set individually, i.e. ‘divide and conquer’. Accordingly, we divide the pose parameters in different sets and estimate each set independently. We provide valid conditions to mask other sets of pose parameters while estimating a particular set. An H-shaped helipad made of a combination of three ellipses is used to estimate all sets of pose parameters. Thanks to simple computations, the system is much cheaper and faster than the current methods in practice. Real-time experiments indicate the effectiveness of the proposed method.