ZigZag Algorithm: Scanning an Unknown Maze by an Autonomous Drone

Abstract

We consider the problem of a drone (quadcopter) that autonomously needs to scan or search an unknown maze of walls and obstacles (no GPS and no communication). This ability (navigating in an unknown indoor environment) is a fundamental problem in the area of drones (even in general robotics) and has applications in military, security, search & rescue and surveillance tasks. Typically, previous works proposed systems that construct a 3D map (via camera images or distance sensors) of the drone’s surroundings. This 3D map is then analyzed to determine the drone’s location and an obstacle-free path. The algorithm proposed here skips over the 3D map and the computation of the obstacle-free path by using random “blind” billiard zig-zag movements to scan the maze. This way, the drone simply bounces from walls and obstacles disregarding the need to find an obstacle-free path in a 3D map. Thus the algorithm requires only a simple form of obstacle detection, one that alerts the drone that there is a close obstacle in its direction of flight. Just using zigzag movements was not enough to obtain efficient cover of the maze were “efficient” cover is when the drone performs no more than one pass per corridor/room (OPTtime). Hence, a more complex algorithm was developed on top of these random zigzag movements. Experimental results using a realistic flight simulation in a random maze showed about 95% cover in OPTtime.