Contact Sensor-based Coverage of Rectilinear Environments

Abstract

A variety of mobile robot tasks require complete coverage of an initially unknown environment, either as the entire task or as a way to generate a complete map for use during further missions. This is a problem known as sensor-based coverage, in which the robot's sensing is used to plan a path that reaches every point in the environment. A new algorithm, CC/sub R/, is presented here which works for robots with only contact sensing that operate in environments with rectilinear boundaries and obstacles. This algorithm uses a high-level rule-based feedback structure to direct coverage rather than a script in order to facilitate future extensions to a team of independent robots. The outline of a completeness proof of CC/sub R/ is also presented, which shows that it produces coverage of any of a large class of rectilinear environments. Implementation of CC/sub R/ in simulation is discussed, as well as the results of testing in a variety of world geometries and potential extensions to the algorithm.