Cognitive abilities of indoor cleaning robots

Abstract

Indoor cleaning robots often work in peopled surroundings. Therefore, cleaning robots in domestic homes should not only clean efficiently but also harmoniously integrate with humans. This paper proposes to add cognitive abilities to intelligent cleaning robots, including environmental cognition, dynamic obstacles, and cleaning tasks. These cognition abilities are implemented by very simple, yet efficient theoretical approaches. First, a robot builds a grid map of its environment via sonar and select human-robot interaction. Then, the robot moves one round along the room boundary via internal spiral coverage (ISC) algorithm to understand the room structure. The cleaning efficiency of a robot is greatly improved by understanding its environment. The structure of a room also provides semantic planning information for task cognition. Moreover, family members are able to avoid a moving robot in different ways. People are therefore divided into three types: elderly people, children, and adults. We assume that people wear different colors of clothing. The elderly wear red, children wear yellow, and adults wear blue. The robot can visually recognize these colors and adopt different obstacle avoidance policies, such as emergent, active, or normal. Finally, three cleaning modes are set: unstructured, structured, and local. Users can select different modes to save energy. Experimental results show that these cognitive abilities of a robot improve efficiency and enhance its safety and perceived friendliness.