Experience based domestic environment and user adaptive cleaning algorithm of a robot cleaner

A common robot cleaner uses complete coverage path planning (CCPP) with a map gotten by simultaneous localization and mapping (SLAM) [1] and it only contains geometrical information. For effective cleaning task, a cleaning robot should consider not only the geometrical map but also domestic properties and a user´s preference. This paper suggests an experience based cleaning algorithm of a robot cleaner. Experiences of concentrated cleaning commands from a user and ones of failures during cleaning because of obstacles which cannot be detected like electric wire are stored as data. They are used to find areas which should be avoided and or should be cleaned with concentration. To estimate the areas´ shape, kernel density estimation [2] is used. This density is used to find property area with threshold. After finding the areas, the robot determines cleaning order among them. As a further work, it will be tested in simulation and be applied to a real cleaning robot which uses SLAM as a next step.