Development of a High Efficiency and High Reliable Glass Cleaning Robot with a Dirt Detect Sensor

From Control Systems Technology Group
Revision as of 21:41, 17 February 2018 by S160587 (talk | contribs) (Created page with '<p>Cleaning of glass windows for buildings is a dangerous work for humans. Thus, automatic robots for glass window cleaning have been developed. There are two major ways for the …')
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Cleaning of glass windows for buildings is a dangerous work for humans. Thus, automatic robots for glass window cleaning have been developed. There are two major ways for the control of the autonomous cleaning mobile robot, which are reaction-based control method and model-based control method. Efficiency of the cleaning using the reaction-based method is lower than one using the model-based method. This study aims to develop a high efficiency and high reliable glass cleaning robot. Thus, the model-based method for the control of the cleaning robot is adopted. For efficient and reliable cleaning, the robots should follow accurately a desired trajectory which covers the whole window with the least possible of the sweep motion while taking into account downwards dripping of water while cleaning the window. Therefore, the robot follows the horizontal parallel motion trajectory in this study. The tracking of this trajectory is hard because of gravitational force and other dynamical effects. These effects should be considered in the design of the control law. Therefore, this paper proposes a new trajectory tracking control method. This control law is a nonlinear feedback one with compensation of gravity and other dynamical effects. For the guarantee of complete cleaning, the robot should have a dirt detect sensor. However, no such sensor has been developed yet. Therefore, this paper proposes a dirt detect sensor including a motion control method. The sensor has a light emitting section and a light receiving one. It detects dirt by measuring amount of reflected light which is basically in proportion to the amount of dirt materials on the glass.

For the trajectory tracking control method, a kinematic level control by Kanayama is used. The control method is then modified to compensate for the dynamic effects. For the control method, it is necessary to know the wheels’ angular velocity and the position and orientation of the window cleaning robot. The angular velocity of the wheels is measured with rotary encoders, the position and orientation are detected by external sensors. Direction of gravity is constantly parallel to the working plane of the cleaning robot on a vertical glass window. Therefore, the real orientation of the robot q can be detected by measuring accelerations along the longitudinal direction and the lateral direction with acceleration sensors. Influences of the moment acceleration are neglected because they are small compared to the gravitational acceleration. Two distance sensors are installed in the robot to detect the real x and y position of the robot with reference to the vertical, respectively horizontal window frame.

Next, the dirt detection method for a glass window is discussed. A possible dirt detection method is to have a light emitting section on one side of the window and a light receiving section on the other side that have a constant distance to each other. This is hard to achieve since the robot has to cope with vibrations generated by the friction resistance of the cleaning unit. Therefore, the dirt detection method that is chosen in this paper is to measure the amount of reflected light which is in proportion to the amount of dirt. This way, the receiving and emitting section can be placed on the same side of the window at the underside of the robot which circumvents the vibration problem. For the dirt detect sensor a line of photosensors is used. These photosensors have a light receiving and a light emitting section and the output voltage that they produce is in proportion to the amount of received light. This line of photosensors is composed of different cells to ensure that highly localized dirt is detected. From every cell, the output voltage is measured before and after cleaning a horizontal strip of the window. These voltages are compared. If they are higher than a certain threshold, the strip needs to be cleaned again, since there is still dirt that can be removed. If not, the robot moves one strip down and repeats the process. This ensures that the whole window is eventually cleaned properly. Using the absolute output voltage of the cells as a measure for the amount of dirt is not an option since the voltage of a cell is also influenced by the variety of the window and its thickness.

At last, several experiment are conducted to check if the proposed trajectory tracking control method, the dirt detect sensor and the motion control method to guarantee the completeness of window cleaning work. The results of these experiments are promising.