Contents
Introduction
COLDLASER is a Research and Manufacturing Organisation that has been set up to deliver ecologically positive AI based laser cleaning systems to both the consumer and the commercial market. For the past four years the founders have been working on a disruptive, ground-breaking platform which combines Drones, Robotics, Artificial Intelligence and our unique low energy Laser technology. Our brilliant tech ecosystem will allow us to develop a suite of world changing cleaning devices.
Our objective to be the largest provider of green consumer laser cleaning products and services globally. We are transforming years of experience creating successful businesses and ground-breaking global projects for clients like Microsoft and Apple into a revolutionary new enterprise.
Background
Inspired by the work of Masahiro Ueda, Ryoya Makino and Ki-ichiro Kagawa of Fukui University, Japan, John Whelan and Po Keung Wong began this journey during the summer of 2019 over a LOT of coffee. We have spent the past few years researching and testing the viability of the concept, the safety and practicality of these devices and of this technology before considering the various routes to market and the next steps. We have determined an initial coin offering to be the most appropriate manner in which to reach out to the types of person/investor that we feel will get our concept and will intrinsically understand what we are trying to achieve.
Market Research
The first challenge we encountered when initiating the Market Research process was the scale of the market. Our fundamental tenet is that we intend to develop the capability to clean any surface built by humans.
Essentially, we intend selling our products into a market in which nothing comparable exists. We are totally disrupting this market and replacing all existing products in current use. We will begin with cleaning glass, specifically, windows. We decided to define the window cleaning market in terms of revenue derived from the process of cleaning windows. This decision was driven by our desire to provide potential investors and believers in COLDLASER with a perspective on the kind of revenues we are capable of generating even with our earliest products.
Our approach has been to begin with a very basic handheld device. Then we create an initial proposed taxonomy of commercial objectives and target markets by utilising the medium through which our Drobots (Autonomous Window Cleaning Robot) travel combined with the method of propulsion to categorise and define each device.
Although the size of the market tends to defy categorisation (for example - how many glass windows are there in the world?) we need to commercialize it to be able to easily illustrate the potential of this ground-breaking technology.
Therefore we shall focus on the first, reduced functionality prototype and in order to define its target market, we reduce this massive figure by asking how many windows can safely be reached by a human being holding our device in one hand? We reduce it further to domestic household windows.
Geographical segmentation further clarifies the target market and we arrive at the following data scenario:
PLEASE NOTE: all statistical data has been sensitized by a factor of 10:1
Geographic Sector A:
United States of America
Population 331.449 million Total households 139.684 million
Average Number of Windows per household 9
Average Cost of Cleaning per household US$75*
Potential Annual Revenue (10%) US$1.048billion
Canada
Population 38.1 million
Total households 15.646 million
Average Number of Windows per CAN household 10
Average Cost of Cleaning 1 x 10 window household CAN$160
Potential Annual Revenue(10%) CAN$250 million
Geographic Sector B:
United Kingdom
Avg Cost Window Cleaning/household €46.74
No of households(million) 27.8 Potential Annual Revenue(10%) €129.94 million
Republic of Ireland.
Avg Cost Window Cleaning/household €60
No of households(million) 1.8
Potential Annual Revenue(10%) €10.8 million
Germany
Avg Cost Window Cleaning/household €56
No of households(million) 41.5
Potential Annual Revenue(10%) €232.4 million
France
Avg Cost Window Cleaning/household €67.50
No of households(million) 28.73
Potential Annual Revenue(10%) €193.93 million
Spain
Avg Cost Window Cleaning/household €35
No of households(million) 18.625
Potential Annual Revenue(10%) €65.19 million
Italy
Avg Cost Window Cleaning/household €32
No of households(million) 25.02 Potential Annual Revenue(10%) €75.06 million
Sector B:
Target Population (million): 327.22
Total Households(million) 115.675
Potential Annual Revenue(10%) €707.31 million
Sector A + Sector B Combined PAR (€): EUR€3,929 million
Xe.com midmarket rate Aug 2021
DATA SOURCES: SEE APPENDIX ONE
Product Concept
PROTOTYPE #1:
A hand-held, window cleaning laser device. It will not require chemicals, cleaning detergents or water. It will not generate any waste other than some light dust particles created during the process which will be dealt with by an integrated fan/vacuum system.
This device will contain a low fluence laser emitter which will clean the surface with a constant speed and beam intensity.
PROTOTYPE #2:
A hand-held device containing laser emitter and dirt detection sensor based on a debris-detection and classification system which utilises a deep Convolutional Neural Network (CNN) approach to clean windows/glass with different speed and beam intensity.
PROTOTYPE #3:
An Autonomous Windows Cleaning Drone Robot directed via a combination of remote human control and a native AI core. This Drobot utilises similar technology as Prototype #2 - handheld device.
The debris-detection and classification system enables our COLDLASER Drobots to identify and remove dirt or debris in different areas of the windows.
The low fluence laser device scans the window as it moves across it and only initiates the laser beam to clean once it encounters dirt or encrustations on the window. This preserves battery life in the device. The device uses hot-swap battery technology.
FUTURE PRODUCTS:
#4. Autonomous Drobot Aeroswarm
cleaning high level buildings and other 3D structures
#5. Autonomous Drobot Cityswarm
cleaning footpaths and roads/streets
#6. Autonomous Underwater Drobot Aquaswarm
cleaning Ship Hulls, Offshore Oil Platforms, Offshore wind/wave turbines etc.
#7. Autonomous Aerospace Drobot Orbitswarm
cleaning satellites and low orbit geosynchronous space structures
Benefits
Cleaning without water, chemicals, waste. Our policy is to use only renewable carbon neutral sources of energy to manufacture our products and to power them.
When we’ve cleaned something - it looks like new!
The Applications are widespread and varied. This is the bit where we get so excited and could just go on for hours but we will limit ourselves to a single, simple example:-
Instead of relocating a large vessel to a drydock in order to careen (clean) it with all of the cost including burning fossil fuels and loss of time associated with this process. Our underwater cleaning Drobot Aquaswarm will clean ships while they are moored in harbour, resulting in massive savings in costs and time to ship operators and owners and new potential revenue streams for Port Authorities and operators.
Technology
LASER RESEARCH & INFORMATION
CLEANING WINDOWS WITH LASERS:
The Laser cleaning process used with glass surfaces is known as Ablation. Normally the energy from the laser passes through the glass substrate and is not absorbed by it. However when the laser beam comes in contact with (primarily) organic and other forms of non-translucent matter such as soot and dirt on the surface of the glass the laser energy is absorbed by this matter and heat is generated. In less technical terms, you can imagine that the layer to be removed is simply vaporized by the laser beam.
CURRENT RESEARCH:
Our current research focus is on Precision Laser Ablation. This refers to the precise removal of thin layers on the target surface which requires tight control of the laser parameters of energy, energy stability and spatial uniformity. When specifying a laser ablation system, it is important to determine the “process window” for the materials to be ablated. This is the energy range which produces appropriate material removal, i.e. acceptable cleaning results.
Energy below this range does not ablate adequately and energy above this range damages or ablates excessive material. The sum of the energy stability and spatial uniformity must not exceed the process window in order to achieve good results. In addition, accurate control of the nominal energy, which can be provided by a digitally controlled attenuator, is critical for predictable results. For these applications, more energy is not necessarily better due to the fact that excess energy decreases the effective resolution.
Hence our name – COLDLASER which refers to the lower intensity end of the energy spectrum inhabited by the laser devices we are creating.
For example, in some glass cleaning experiments the best cleaning results were obtained irradiating with one pulse at the fundamental wavelength of Nd:YAG laser (1064 nm, pulse duration of 6 ns) with fluences of 100 - 600 mJ*cm-2. Moreover, when the cleaning proceeded on the colourless glass, the process was self limiting, due to the glass transparency, to 1064 nm. Cleaning using wavelengths at 532 nm required the application of many laser pulses and higher energies of laser beam.
Over the past three years our target has been to identify the correct Process Window for the laser technology we intend to use. The correct Process Window will achieve two key things:
1. Optimal Cleaning range to ensure efficient cleaning of the target surface.
2. Maximum Laser Power Requirement which will determine the size and shape of the equipment we will build.
SENSOR TECHNOLOGY:
Initially all AI cleaning robotics and drone products (collectively described as Drobots) designed by COLDLASER will utilise built-in advanced Lidar Technology. In the case of our AI cleaning Drobots, a spinning lidar sensor will be integrated on top of the Drobot that maps out the environment/road prior to initiating cleaning. This sensor fires a laser in every direction; then, the laser light reflects back into the sensor to collect the data that helps the Drobot map out the environment/road before performing any cleaning action. This data is stored on the Cardano Blockchain.
The LIDAR sensor on each AI cleaning Drobot scans the environment/road in milli-seconds and from this our AI creates the most efficient cleaning pattern .
Combining the real time map built by the LIDAR sensor with existing mapping data, each cleaning Drobot divides environmental areas into sections and moves around in straight lines for the most efficient path possible.
Our AI cleaning Drobot Cityswarm cleans all environmental/road surfaces—Streets, footpaths, etc.
Each AI cleaning Drobot will map out their own paths while avoiding any obstacles.
BACKGROUND INFORMATION:
1.1 Dry Laser Cleaning
Dry laser cleaning is widely used in industry and researched in academia. It is generally based on a rapid thermal energy transfer between the incident laser beam and the substrate/contaminant. The resulting rapid thermal expansion of the components provides the force to remove the particles.
(Fig. 1.1)
Figure 1.1. Mechanism of dry laser cleaning.
1.2 Angular Laser Cleaning
In contrast to typical laser cleaning, the angular laser cleaning technique irradiates the contaminant surface at a glancing angle (Fig. 1.2). This is claimed to drastically improve the cleaning efficiency by 10 times compared to typical laser cleaning that uses perpendicular irradiation. However, its effectiveness is dependent on the surface characteristics and the size of the contaminants.
Figure 1.2. Mechanism of angular laser cleaning.
1.3 Common Laser Systems Used for Laser Cleaning
Depending on the materials, applications, and required speed, various laser sources are used for cleaning. Table 1.3 presents some of the different types of laser systems used for cleaning along with the wavelengths and their typical usage.
Figure 1.3. Common Laser Systems Used for Cleaning
1.4 Typical Contaminants Found in Industrial Components
This section describes the typical contaminants that are present on the surface of typical industrial components. Most of them may be deposited during movement of a workpiece in the industrial environment. The typical contaminants that may be present on the surface of the industrial components include:
• Hydraulic oils
• Silicone greases
• Machine tool coolants
• Organic and inorganic particles
• Metallic swarf
• Native surface oxides
Some few contaminants are often bound very strongly to the surface, making their removal difficult. The adhesion of contaminant particles to a solid substrate depends on secondary bonds. Secondary bonds are weak in comparison to primary bonds (ionic, covalent, and metallic). In dry surroundings the main forces of adhesion to solid surfaces are: (i) electrostatic, (ii) secondary van der Waals forces, and (iii) hydrogen bonding.
Electrostatic action is generated by electric charges, which form at the contact interface between two materials. Van der Waals forces are generated by dipole interaction between two molecules. A dipole can be induced on a molecule by the permanent or instantaneous dipole moment of another molecule, thereby creating a force of attraction between the two molecules. Therefore, there exists a weak attractive force acting between all atoms and molecules at all times, allowing molecules to stick together even when there are no attractive forces due to permanent charge separation in the molecules themselves. A hydrogen bond is formed by electrostatic attraction between a hydrogen atom bound to an electronegative atom and another electrically negative atom.
In general, contaminants adhere to the material surface by one or another of these secondary bonds. The main force of adhesion of small particles, less than 50 μm diameter, on a dry surface are van der Waals forces. Electrostatic forces become important and predominate for particles larger than 50 μm diameter. Electrostatic forces also play an important role in attracting particles to the surface for adhesion. The total adhesion forces acting on a particle of only 1 μm diameter can exceed gravitational forces acting on that particle by a factor greater than 106. In a humid atmosphere, other forces due to capillarity and surface tension will be active. The problems of cleaning can be considered as comparisons of the strength of adhesion between the dirt and the object being cleaned and the cohesion of the molecules of the object to one another.
1.5 Dry Laser Cleaning Method
The DLC approach was the first method employed to remove particles from substrates using lasers. Various academic and industrial research groups have reported the experimental, characterization and computational results of this cleaning method in the past. In the dry laser method of cleaning, a short-pulsed laser beam is directed on the substrate that has to be cleaned. This laser pulse excitation on the substrate results in rapid thermal expansion and thermomechanical wave propagation and out-of-plane acceleration, thereby exciting the substrate and/or the particles that have to be removed. This high-frequency (nanosecond) acceleration and wave propagation phenomenon generates an inertial force that can shake off the particle adhered to the substrate, provided the generated inertial force exceeds the total adhesion force consisting of several individual forces, such as van der Waals, electrostatic, and capillary forces. The main principle of this mode of sub-micrometer particle removal is the substrate acceleration induced by the thermoelastic field generated by the irradiation of the short-pulsed laser. The substrate attains its maximum acceleration value due to the substrate thermal expansion during the irradiation of the short-pulsed laser in the out-of-plane direction until the peak fluence of the beam is reached. The resulting positive acceleration presses the particle down and increases the contact diameter and the strain energy stored in the deformed particle. Removal can occur only after the surface begins to decelerate. The magnitude of surface acceleration is proportional to the level of fluence due to linearity assumption of thermoelastic effects and is inversely proportional to the duration of the laser pulse squared. However, above a certain level of laser fluence, thermal and/or mechanical damage on the surface could occur. It is also noteworthy that during this process the surface is subjected to high levels of electromagnetic radiation as well.
In the dry laser cleaning technique, to understand the modes of substrate damage, both thermal and mechanical (stress) damage thresholds must be well understood and accurately modelled. A key complication in such analysis is that the material properties vary substantially with strain rates and temperature changes, and it is often not well understood how certain materials yield under high thermoelastic strain rate excitations. These thresholds could then be used to avoid excessive heat deposition and/or stress levels. The complexity of the thermoelastic process requires that a detailed analysis be carried out for the determination of optimal removal efficiencies.
It was reported that for the DLC method the required laser fluence for removing a particular size particle depends on the coefficient of reflectivity of the substrate and the particle. Based on the computational model, the critical limit of DLC was determined, in terms of the minimum diameter of the silicon particles that could be removed from silicon and copper substrates without any damage. The research group at the Photo-Acoustics Research Laboratory at Clarkson University has reported that the type of damage initiation (thermal or mechanical fracture) depends on the particle size that has to be removed and particle-substrate system material properties.
The thermal damage threshold is the melting temperature, and mechanical damage occurs above the yield stress of the substrate material. It should be noted that the only effects considered in the reported study were linear thermal and mechanical fields, and no other types of damage such as optical and electromagnetic damage mechanisms were considered. Delicate structures on substrates and processing techniques used to machine them sometimes substantially lower the local material yield properties, and such structures can amplify the laser beam on the textured surface due to diffraction, thus their cleaning requires substantially greater care than flat substrates.
1.6 LASER ABLATION
The laser ablation process covers a great diversity of particular applications in which mass removal of laser irradiated materials is the essence of the final required process. Even more than previously mentioned cutting and welding processes, physical mechanisms involved in laser ablation are extremely complex (Von Allmen 1987) depending on the particular ablation technique considered (molten material ablation, vaporization phase ablation, sublimation techniques, non thermal ablation, etc.). Quality control in laser ablation implies surface final state characterization, including, if possible, estimation of ablated mass, walls morphology in ablation fronts and layer behaviour in multilayer laser ablation. That is also the case in laser cleaning (Meja 2001), and other processes as laser lithography (Lamda Physics 2001) and surface modification (Kaplan 1998) . Even for fully commercial laser ablation applications such as laser marking and engraving CLSM offers complete assessment including marking depth, walls slope, pattern homogeneity, etc. (Fig. 6).
Figure 1.6 Laser Ablation
1.7 Methods for Monitoring and Measuring Cleanliness of Surfaces
Laser-Induced Breakdown Spectroscopy
LIBS is a minimally destructive, atomic emission spectroscopic technique in which successive nanosecond laser pulses ablate a small amount of material from the surface of interest. The resulting micro plasma is used to interrogate a material (solid, liquid, or gas). Excited gas phase atoms, ions, and molecular fragments formed within the plasma, emit fluorescent radiation that is characteristic of the material being sampled. If the same point on the surface is ablated successively, the LIBS spectra can be collected to provide in-situ cross-sectional analysis of the material. Raman microscopy can achieve very high spatial resolution (~1 μm) and gives spectra largely free of contamination from surrounding material. By combining these techniques, in-situ elemental analytical data and chemical composition of the contaminant of interest can be obtained. The combined LIBS–Raman technique has been used for identification and analysis of contaminants in pigment in artworks. For example, the original white paint on a Byzantine icon was analysed to be a hydrated lead carbonate (2PbCO3.Pb(OH)2), but the subsequent restorative work used zinc oxide paint. LIBS can be combined with laser cleaning applications or other conventional cleaning methods for monitoring and controlling the inspection and cleaning process for culturally relevant structures and artwork. LIBS analyses have been performed at longer distances (50–100 m) for various applications, such as hazardous material detection and industrial monitoring. The LIBS technique can be used to obtain information on the size, number density, mass and composition of a wide range of contaminant particles, including metal hydrides, coal particles, halons, pigments, and single elements, such as As, Be, Cd, F, Fe, Mn, Ni, Pb, and Hg.
Particles as small as 150 nm have been analysed, corresponding to an absolute detectable mass of 10−15 g. This shows the very high sensitivity of the LIBS technique for particle characterization.
LIBS features several key advantages for the analysis of surface contaminants in industrial applications, as well as on cultural heritage objects and artworks. The analysis can be performed in-situ and only requires an optical contact with the object. Its ability to detect light elements, such as Be, and having very low detection limits for many elements is an advantage over other in-situ techniques, such as X-ray fluorescence. The technique does not require sampling, and little or no sample preparation.
Furthermore, LIBS is a very rapid technique (limited by readout speed of the detector), as the information is recorded with a single laser pulse measurement. The technique is very slightly destructive, as the material ablated from the surface is minimal. In addition, LIBS allows depth profiling of a surface by applying successive laser pulses on the same spot. The equipment required to perform LIBS is technologically mature, and a variety of commercial systems are available. The equipment can be easily ruggedized. A disadvantage of LIBS for in-situ measurements is that the surface concentration may be below the instrument detection limit. This may be a serious issue because of the small area sampled by each laser pulse (<1 mm2).
Another recent innovation has been to combine LIBS with SNOM to map and correlate elemental chemical compositions of surfaces with the surface topography. In this method, surface topography is mapped by scanning the surface with a SNOM probe. The probe is then positioned over the feature of interest, such as a contaminant particle, that is ablated with a normal or ultrafast laser pulse. The LIBS spectrum of the feature is obtained from the optical emissions resulting from the micro plasma plume.
We are grateful to the following Scientists and Academic figures whose wonderful work we utilised to guide our research over the past 3 years, samples of whose findings we quote above to illustrate the technological requirements and complexity of the platform we are constructing:
· Sundar Marimuthu, …Alhaji M. Kamara - Developments in Surface Contamination and Cleaning: Applications of Cleaning Techniques, 2019
· M.D. Murthy Peri, ... Cetin Cetinkaya - Developments in Surface Contamination and Cleaning: Methods for Removal of Particle Contaminants, 2011
· C. Molpeceres, ... J.L. Ocaña - Recent Advances in Multidisciplinary Applied Physics, 2005
· Rajiv Kohli - Developments in Surface Contamination and Cleaning: Detection, Characterization, and Analysis of Contaminants, 2012
The ICO
COLDLASER will launch RAYX TOKENS on the Cardano Blockchain.
Why did we select Cardano?
We need 100% certainty that all commands executed by our Drobot operating system and all 3-D location points visited by our Drobot cleaning devices are recorded securely on an OS Blockchain. In addition, it is vital to have total control at all times over what is virtually an autonomous device and thereby ensure the safety and security of any human beings, animals or structures that might enter or exist within the boundaries of the spatial area defined by the 3-D map created by the programmed location points that our devices will visit when in operation. The languages used in Cardano include Haskell, a statically-typed, purely functional programming language which is extensively used within the Nuclear Power industry.
We propose to issue RAYX tokens in the following tranches and stages.
PHASE 1 - MAIMAN
VOLUME: 86,000,000 RAYX
VALUATION: 1 RAYX = 0.50 ADA
DURATION: MAY 15th 2023 to MAY 21st 2023
MARKET CAP: 43,000,000 ADA
PHASE 2 - TOWNES
VOLUME: 109,382,596 RAYX
VALUATION: TBC
DURATION: NOV 2024
MARKET CAP: TBC
PHASE 3 - SCHAWLOW
VOLUME: 138,109,338 RAYX
VALUATION: TBC
DURATION: SEPT 2025
MARKET CAP: TBC
PHASE 4 - FIOCCO:
VOLUME: 167,454,937 RAYX
VALUATION: TBC
DURATION: SEPT 2026
MARKET CAP: TBC
The TOKEN
Please note that the RAYX token is NOT a tokenized stock, options token, insurance-based token or any form of synthetic asset. It is a generated and not minable token. The total amount of RAYX Tokens that will be generated is 2,000,000,000.
The RAYx Token is a Utility Token, ownership of which will entitle the holder to either a 50% Discount or the purchase at cost (whichever is the lesser) of PRODUCT #1 and PRODUCT #2. We regret that the RAYx token is not available for sale to citizens of the United States.
ROADMAP
2019 – Initiation of Project.
R&D Phase Commences
2021 – R&D Phase Continues
Project Validation
2022 – ICO Preparation
R&D Continues
2023 – MAY - PHASE 1 - MAIMAN
RAYx TOKEN LAUNCH
86,000,000 TOKENS
DEVELOPMENT OF PRODUCT PROTOTYPE #1
2024 – NOVEMBER - PHASE 2 - TOWNES
MANUFACTURE & SALES OF PRODUCTS #1
DEVELOPMENT OF SWARM AI Operating System
2025 - SEPTEMBER - PHASE 3 - SCHAWLOW
DEPLOYMENT OF SWARM AI OS
2026 - SEPTEMBER PHASE 4 FIOCCO
Use Of Funds
PHASE 1 FUNDS TARGET - €17,200,000
Recruitment & Salaries
Lab & Office
Legal/Professional/Patent extension work
Hardware & Software
Manufacturing & Trials
PHASE 2 FUND TARGET - TBC
MANUFACTURE/SALE OF PRODUCTS #1
CLIENT FACING
Revenue Team
Domestic/Retail
SDR & AE Teams
Utility/Government
SDR & AE Teams
Corporate
SDR & AE Teams
Customer Relations Team
Marketing
Manufacturing
Logistics
As Phase 1 above
PHASE 2 R&D
PHASE 3 FUND TARGET - TBC
CLIENT FACING
Revenue Team
Domestic/Retail
SDR & AE Teams
Utility/Government
SDR & AE Teams
Corporate
SDR & AE Teams
Customer Relations Team
Marketing
Manufacturing
Logistics
As Phase 1 above
PHASE 4 FUND TARGET - TBC
CLIENT FACING
Revenue Team
Domestic/Retail
SDR & AE Teams
Utility/Government
SDR & AE Teams
Corporate
SDR & AE Teams
Customer Relations Team
Marketing
Manufacturing
Logistics
As Phase 1 above
Team
FOUNDER – Po Keung Wong 9.15% RAYX)
FOUNDER – John Whelan 9.15% RAYX)
MANAGEMENT TEAM
Strategic Advisor Team
We believe that it is vital that there are checks and balances within the decision making process of COLDLASER and in the interests of good governance, we are in the process of establishing a group of external, neutral advisors to help steer the company and keep us focussed on our drive towards constant innovation and product improvement. A minimum of one seat in this group will be an elected member of the RAYX token Community.
PRIVACY TERMS
FAQs
Appendix 1
MARKET RESEARCH Datasources:
United States Census Bureau Population Census April 1 2020
United States Census Bureau Housing Units Est. July 1 2019
United Stages Department of Energy DOE/EE-0965 • September 2013 (Avg. no. of windows)
https://homeguide.com/costs/window-cleaning-prices - 2021
https://www.fixr.com/costs/window-cleaning
*Note: fixr.com state that the minimum average window cleaning cost across the USA is $75 to $150. In the interests of prudence, we have selected the lowest end of the range - $75.
Statistics Canada Population Estimate April 1 2021
Statistics Canada Household Statistics June 26 2020
https://www.bark.com/en/ca/window-cleaners/window-cleaning-cost/
https://www.checkatrade.com/blog/cost-guides/window-cleaning-prices/
https://dublin-windowcleaning.ie/window-cleaning-prices/
https://www.howtogermany.com/pages/domestichelp.html
https://www.vitrissimo.fr/espace-particulier/tarifs-des-prestations/tarifs-nettoyage-vitres-a-domicile/
https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/families/bulletins/familiesandhouseholds/2020
https://en.wikipedia.org/wiki/List_of_countries_by_number_of_households
https://www.destatis.de/EN/Themes/Society-Environment/Population/Households-Families/_node.html
https://www.vitrissimo.fr/espace-particulier/tarifs-des-prestations/tarifs-nettoyage-vitres-a-domicile/
https://www.ined.fr/en/everything_about_population/data/france/couples-households-families/households/
https://www.ine.es/en/prensa/ech_2019_en.pdf
https://en.wikipedia.org/wiki/List_of_countries_by_number_of_households