1.1 - Project Proposal Spring 2026
Window Washer
Introduction:
Did you know that it takes a team of 36 window washers 3 months to clean the windows of Burj Khalifa, the tallest building in the world? Moreover, the cleaning process is repeated four times per year, so the cleaning is essentially nonstop. These people serve a critical purpose in keeping the windows clear and shiny for the people inside and outside of the building. However, the profession can not only be quite harrowing, but workers can also face dangers, including equipment malfunctions, environmental factors such as wind, rain, heat, and storms, and chemical exposure from cleaning solutions. Moreover, with aging window cleaners and a growing labor shortage, we need to consider alternative solutions to ensure that high-rise windows are cleaned while also helping improve worker safety.
[Source: CNN]
Therefore, we wanted to take some inspiration from the fact that robotics companies have started developing mechanisms that can automate window washing. This problem is very practical and interesting to solve. We would need to consider a wide variety of factors, and therefore this project would be a great opportunity to explore forces (gravity), controlling position and motion throughout the descent/ascent, while actively cleaning the actual window itself, using a four-bar linkage, and minimizing the number of actuators.
[Source: The Robot Report]
Problem Statement:
Designing a window washing mechanism presents several technical challenges. The challenge involves managing vertical positioning while maintaining contact with the window and cleaning the entire area. Additionally, there is the added challenge of actuating several components with a single actuator.
To manage the vertical positioning, the mechanism, supported by two lines, will need to have precise position and motion control while dealing with external forces (mainly gravity). Since the goal is to ensure that the entire area is cleaned before it descends, we will need to create some form of intermittent motion that allows for a full cycle of the cleaning mechanism before the mechanism descends just enough to begin cleaning the next area. Moreover, given the force of gravity, the mechanism will need to be carefully designed to keep a low and centered center of gravity while also supporting its own weight. Lastly, during periods of non-descent, the mechanism must remain stable.
We also need to carefully consider the cleaning portion of the mechanism. While the mechanism is hanging on two ropes, the cleaning pads need to stay in contact with the glass at all times. Moreover, the cleaning mechanism needs to be able to cover a decently vast area, involving some sort of reciprocating motion.
The design of this mechanism needs precise control of position, velocity, acceleration, and external forces. This cannot be accomplished with simple joints, as the mechanism must carefully manage its intermittent descent with continuous cleaning, all with the use of a single actuator.
Mechanism Description:
Our plan for the mechanism will be a combination of a four-bar mechanism, specifically a slider-crank mechanism, and a Geneva mechanism for intermittent motion.
The mechanism will be suspended from above using two strings, one on either side, attached at the top of the window using one or two strong suction cups. For controlling vertical positioning, we will have two pulleys, one on either side, that will be attached to the strings themselves. These pulleys will be mechanically connected to the intermittent output side of the Geneva mechanism. This allows for a controlled descent/ascent while having continuous motor output. Additionally, both pulleys will need to be synchronized with each other, likely via gears and belts/chains, to ensure both sides descend/ascend at the same rate at the same time.
The Geneva mechanism will convert the continuous motor input into intermittent motion, allowing the cleaning mechanism to complete a full cycle before changing position vertically. The Geneva mechanism geometry should help prevent unintended downward motion during the stationary periods. Additionally, we may use ratchets to prevent unintended descent during stationary portions of the cycle and help the mechanism remain stable while suspended.
Additionally, we plan to have one slider-crank mechanism with a squeegee attached to clean the glass. This slider-crank will have continuous motion powered by the same actuator as above. We may use spring-loaded ends or gravity to maintain sufficient contact force between the cleaning pads and the glass throughout the cleaning cycle.
The slider-crank and Geneva mechanism positions must be carefully designed so that their cycles are interconnected. To maintain balance, we need to ensure that the center of mass remains fairly centered throughout all positions and movements
Scope of Work:
The scope of this project is to design, analyze, and build a window-washing mechanism capable of producing both continuous cleaning motion and intermittent vertical movement by the end of the semester.
First, we will perform kinematic analyses to determine the motion requirements of the mechanism. We will begin with positional analysis to determine the cleaning stroke produced by the slider-crank mechanism and the amount of rotation produced by the Geneva mechanism each cycle. Using this positional analysis, we can determine the amount of vertical displacement each cycle, along with how much area is cleaned and how much of the cleaned area overlaps each cycle. Next, we need to perform velocity and acceleration analysis to ensure not only a smooth, reasonable cleaning cycle, but also to ensure balance as the pulleys are turned. Force analysis will also be required to determine whether the actuator provides sufficient output torque, whether additional gearing is needed, and whether the Geneva mechanism can resist motion during stationary periods and thus support the weight of the mechanism. We also plan to animate the linkage motion in Python so that the position and velocity of the mechanisms can be visualized throughout the full cycle.
Before fabrication, a combination of CAD and prototyping will help inform our overall design. Modeling the mechanism fully will help us visualize packaging and identify any possible interference. Additionally, it will allow us to visualize the actuation from the actuator, through the Geneva mechanism, to the gears, to the pulleys. Prototyping will be extremely important as we explore the torque of our motors and the geometry of the Geneva mechanism, which will allow us to decide whether we need additional gearing and/or a ratchet mechanism to prevent undesired downward motion due to gravity. Additionally, we will explore different approaches to keeping the squeegee against the window during prototyping. After the initial CAD phase and prototyping, we will create a full CAD model that will be used for fabricating parts via 3D printing and laser cutting.
We are currently still deciding whether the final project will focus on controlled descent or controlled ascent (controlled descent means less output torque is required and is more in line with a real window washer, but controlled ascent is more interesting because we are going against gravity). As stretch goals, we may attempt both controlled ascent and descent while also improving the coordination between cleaning coverage and vertical displacement so that the squeegee stays at the same height while the robot moves downward, ensuring smooth, separated areas of cleaning during each cycle.
Preliminary Designs:
Kinematic Diagram + Gruebler-Kutzbach Calculations:
Links from drive → output, treating Geneva mechanism as a modified gear full joint:
Ground: L1
Motor Pulley: L2 (g: O2)
Geneva Driver via belt: L3 (g: O3, h: A23)
Slider Coupler via bearing: L4 (f: B34)
Slider Block via bearing: L5 (g: O5, f: C45)
Geneva Follower via bearing: L6 (g: O6, h: D36)
Right Pulley via belt: L7 (g: O7, h: E67)
Left Pulley Gear via gear: L8 (g: O8, h: F68)
Left Pulley via belt: L9 (g: O9, h: G89)
Key: g = full joint to ground, f = full joint, h = half joint (belts/gears only remove 1 additional DoF)
9 links, 2 full joints between components, 7 full joints to ground, 5 half joints (belts/gears) between components = 1 DOF
Preliminary CAD: