The Product

What it is

We’re making a drone capable of cleaning high-rise and skyscraper windows

How it works

Cleaning

We understand that window cleaning professionals take pride in their work and won’t settle for a second rate clean. That’s why we’re investigating several methods of window cleaning to find a solution that meets your standards. We’ve made preliminary steps toward mechanising a water-fed pole, looked into the traditional squeegee, and researched advanced cleaning technologies.

Our main research focus currently is on ultrasonic cleaning, a highly effective technology for removing dirt and grime at the microscopic level. While typically ultrasonic cleaning requires submerging an item in a solution bath, some advancements have been made toward adapting this cleaning method for vertical surfaces. See the following video from the Tekniker Institute for more information: https://www.youtube.com/watch?v=9_q05RO1yzM

Flying

In the last decade quadcopters (commonly referred to as “drones”) have become increasingly popular. Supporting this growing interest is a large community of developers who contribute to the software that allows these machines to fly. This software has become more and more advanced, enabling stable, controlled flight in tight spaces. We are leveraging this software, much of which is open source, and making proprietary adaptations to meet our specific requirements.

Control and Navigation

Robots are powerful, but will always require human supervision. Our goal is to augment the ability of professionals to work quickly by providing an efficient system that’s easy to use. To aid us in doing so we plan to build as much automation as possible into early product releases and then afterward continue improving that software and building in more automation. For example, with early releases a human pilot may need to to manually navigate the craft to a window, but once there the system can be commanded to execute an automatic cleaning sequence. Eventually we hope to use AI algorithms to teach our system how to recognize a window, and SLAM algorithms so that it can find them by itself.

Problems We Face

Weight and Power

Operating a multicopter non-stop takes a lot of power. Most drones can’t stay airborne for much longer than 20 or 30 minutes, let alone several hours. To get around this issue current iterations don’t carry their own battery, but instead have a power-line running down to them from the top of a building. Carrying water is also difficult so we intend to run a waterline as well. If current research pans out however, we can use significantly less water. That combined with an alternate form of energy storage means that hopefully we will be able to cut the cord.

Wind

You get up high enough and it gets windy. We know that. Fortunately, tall buildings tend to break the wind, so while it’s still blustery at least it’s less of an issue where we’ll be flying. This problem is something we are looking into and have good reasons to believe that we will be successful in solving. As soon as we have a prototype we feel comfortable testing on at height, we’ll be working on this issue.

Control

Software gets complicated really fast. The drone needs to stay stable up next to a building and it can never lose control and fall to the ground or into a building. At the same time, the drone needs to clean efficiently. Drones have not been used in many applications where they come into contact with an object. This means that we have to consider what can happen to the drone and how to control it effectively. This is at the forefront of our current efforts.