Reliable Glass Cleaning Motion of a Wall Climbing Robot Using a Dirt Detect Sensor

Abstract

The paper deals with the problem of automatic cleaning of glass windows. A small-sized glass cleaning robot has been proposed by authors. The robot is composed of two units sandwiching the glass by magnetic force. To improve the functional performance of the robot and increase the cleaning reliability, a sensor detecting the level of the glass dirt is introduced. The sensor is composed of light emitting and receiving sections. It detects the dirt by measuring the amount of reflected light which increases in dirty parts and decreases in clean parts of the glass. A signal processing scheme for this sensor is presented, and a motion control method incorporating the dirt sensor measurements is constructed. The validity of the control method and the feasibility of reliable cleaning are tested under experiments.