Simulation and Experimental Evaluation of Complete Sensor-based Coverage in Rectilinear Environments

Abstract

Although sensor-based coverage is a skill which is applicable to a variety of robot tasks, its implementation has so far been limited, mostly by the physical limitations of traditional mobile robots. This paper presents sensor-based coverage algorithms both for a single robot and for a team of independent robots which have been designed to allow for easy integration on to real robots. The specific robots in question are planar robots called couriers, components of the minifactory, an automated assembly system. The couriers have excellent position sensing, which enables them to perform coverage, but have no explicit range or contact sensors to detect boundaries, which adds to the complexity of the coverage algorithm. A set of experiments from simulation is presented to show the overall efficiency of the single-robot and cooperative coverage processes in a variety of environments. A second set of experiments performed on a real robot demonstrates the ability to reliably perform sensor-based coverage and also illuminates the effects of specific choices in the type of control used.