Smart glass and artificial intelligence

Contents 1 Introduction 2 Use cases 2.1 Security threats in large cities 2.2 Theft of high-value items 2.3 Reducing light damage in museums and luxury retail 3 What is smart glass? 4 Artificial intelligence 5 Can smart glass 'learn'? 6 ArtRatio 7 Case studies 7.1 The Waterloo Map 8 What does the future hold for smart glass? 9 Author Bio

[edit] Introduction

Smart Glass is only as smart as the control system that drives it.

Whether it is for viewing the exterior world from within buildings or vehicles, or filtering incoming daylight onto interiors, smart glass systems must now evolve to a position where they are continuously learning from their environment, adapting to changes in local and global events.

For this, the future of smart glass systems must reside with artificial intelligence.

Your smartphone can adjust its behaviour based on your personal data and local events:

• dimming down the screen when the battery is low,
• detecting the risk of a heart attack using a medical app, and
• finding the best sushi restaurant, near you, right now.

Similarly, a smart glass system should also be able to adjust its behaviour based on what is happening in its local environment and what it is learning from global events as they unfold in real-time.

[edit] Use cases

We have seen some excellent examples in the news recently from smart glass manufacturers such as View and Halio in which proprietary artificial intelligence (AI) algorithms adapt building facades to changes in sunlight, with clear benefits for sustainability, privacy and comfort.

Let us now consider some more advanced use cases that would take this a step further:-

[edit] Security threats in large cities

Just as car doors unlock automatically on detecting an accident, smart glass facades could also default to a 'safe' transparent state in emergency situations.

For example, if an AI law enforcement app monitoring social media chatter detected an imminent terrorist threat in a large city during a bright sunny day, it is probable that any building with a smart glass facade would be in its energy-filtering ‘dark mode’, to minimise glare and air conditioning costs. This dark mode however would also impair visibility into the building for law enforcement and security agencies trying to ensure everyone has been evacuated. In this case, smart glass facades in the immediate vicinity of the security threat could automatically transition to an ‘emergency mode’ in which they are held indefinitely in a transparent state.

This same principle could also be applied to smart glass systems installed in transportation, enabling a visual line of communication with passengers trapped after an accident, or when the vehicle predicts the risk of a stranded infant or pet suffering possible dehydration.

[edit] Theft of high-value items

Whether in a private residence, a high-end jewellery store or an art gallery, if smart glass is being used in external windows, interior partitions or in showcases, and if the AI algorithm monitoring CCTV cameras anticipates an imminent theft, the smart glass system could ‘go dark’, obscuring the items and slowing down the attacker from finding the most valuable pieces.

When the premises are closed, security guards still need to be able to see the displayed items while on patrol. However, if the building security is likely to be breached, the sensible solution would be to hide everything from view. Not just the most valuable items, but everything.

[edit] Reducing light damage in museums and luxury retail

Global art and luxury retail inventories worldwide can amount to billions of dollars in assets, but excessive light exposure can cause severe damage to paintings, rare books, fashion textiles and even fine wines. If the smart glass control system ‘knew’ which materials were on display (and their sensitivity), it could dim the glass.

This change in illuminance would be practically imperceptible to humans since the eye is an approximately logarithmic sensor (Weber–Fechner law), but it would substantially reduce the long term damage to artworks and luxury materials, since light exposure is the time-integral of the light level. i.e. we calculate light exposure by multiplying the light level by the exposure time.

Future smart glass systems should also learn to automatically identify which items are most vulnerable to damage using AI-enabled machine vision, adjust their behaviour proactively and intelligently to preserve the value of the items displayed.

[edit] What is smart glass?

Smart glass is not a product, but rather a family of technologies.

Broadly speaking, this family can be divided into two major groups: ‘passive’ and ‘active’. In both cases, a stimulus triggers changes in certain properties of smart glass, such as its transmittance, its reflectance, its ability to refract light (i.e. scatter light) and even its electrical conductivity.

• Passive smart glass reacts to environmental stimuli such as light or temperature. Examples include thermochromic and photochromic smart glass which dims in response to heat or ultraviolet light (respectively). This ability to dynamically reject various spectral portions of solar radiation can reduce air conditioning costs in buildings and vehicles, reduce glare for office workers and reduce damage to interior furnishings and retail inventories.
• Active smart glass on the other hand is driven electrically, triggered by manual switches, sensors or control commands from a building management system (BMS).

Examples of active smart glass technologies include:

• suspended particle devices (SPD-SmartGlass), patented by Research Frontiers Inc.
• electrophoretic smart glass technology, such as eLstar Dynamics
• liquid crystal smart glass such as Merck Eyrise or Smartglass International (note: when the liquid crystals are dispersed in a polymer film this technology is called ‘PDLC’ smart glass),
• electrochromic glass (examples include View, Halio and SageGlass) and
• micro-blinds, created by the National Research Council of Canada (NRC).

Within the smart glass family we could also reasonably include ‘transparent photovoltaic’ (TPV) glass, which converts part of the electromagnetic spectrum into electricity (albeit not as efficiently as solar cells). Examples include products from Ubiquitous Energy and Clearvue Technologies.

The advantage of TPV smart glass over solar cells is that the former is transparent to visible light, thus allowing its use as 'electricity-generating windows' in exterior building facades. When incorporated directly onto the building facade, it is often referred to as 'Building-Integrated Photovoltaics' (BIPV).

Other materials which can be classified within the ‘smart glass family’ include smart mirrors, augmented-reality spectacles (think Vuzix, RealWear, Google Glass or the much-anticipated Apple Glass) and heads-up displays (HUDs) for transportation.

[edit] Artificial intelligence

Artificial intelligence (AI) aims to predict future outcomes based on historical data, rather than relying on pre-programmed rules.The system is ‘trained’ to recognise patterns, behaviours or properties and, over time, learns what response to execute. Artificial intelligence is often seen as an umbrella term, encompassing machine learning, which can be specialised to deep learning and further specialised to neural networks.

Current examples of AI applications include automated stock trading, customer service chatbots, Netflix’s recommendation engine and self-driving cars. The key aspect of AI systems is that they adapt in real time, as opposed to traditional systems which are pre-programmed to obey a static set of rules.

AI algorithms could realistically drive smart glass facades or products more intelligently, and even pre-empt certain dangerous or undesirable circumstances.

[edit] Can smart glass 'learn'?

Smart glass itself is just a material, like concrete, stone or wood, with the key difference being the change in its optical properties when driven by a stimulus. When coupled with an AI system, the whole ‘smart glass system’ could learn, but this depends on whether there is a dataset to teach the system the difference between 'good' and 'bad'. And this depends on whether there are sufficient real-time sensors recording the events which feed this dataset.

Based on the use cases outlined above, these ‘sensors’ might include:

• AI bots that monitor social media chatter, alerting law enforcement agencies to increased risks of a threat to the public;
• AI computer vision systems that recognise a known felon in the vicinity of likely targets of theft;
• AI light exposure monitors that predict an increased risk of damage to light-sensitive materials such as art, textiles, pharmaceutical drugs, as well as phototoxic oils, perfumes and dyes.

[edit] ArtRatio

At ArtRatio, we are not yet using AI but rather a modified version of a PID (Proportional–Integral–Derivative) algorithm, similar to what you would find in a cruise control system in a car. Our algorithm is adaptive. It is real-time. But it is not AI (yet). It is sufficiently novel and ‘non-obvious’ though, that we were granted a European Patent for it in 2021.

Here is the reason why: If you subject an unpainted stone sculpture to light levels of say 500 Lux, it will not suffer noticeable deterioration. However, if you put a 15th century silk antiquity under those same light conditions, it can deteriorate in a matter of days. The inventories of museums, auction houses, corporate art collectors or luxury retailers can amount to hundreds of billions of Dollars, and this value is literally being erased by light exposure.

However, not all objects are the same. For this reason, ArtRatio smart glass vitrines modify their behaviour in real time based on what is being displayed, as well as local environmental variables and visitor engagement.

Some materials can be sensitive to light, but others are sensitive to temperature or humidity. Since about 50% of solar radiation is infrared (i.e. heat), incoming sunlight can increase the local temperature inside a display case, which decreases the relative humidity in the enclosed space. So, you see: light, temperature, humidity and even the electrical conductivity of air are all connected, with the damage often being driven by incoming light. And since active smart glass is an electrically-controllable light filter, it seems obvious to use it to do just that, whether in the form of a display case, or by installing the smart glass on the building facade itself.

[edit] Case studies

[edit] The Waterloo Map

During the course of this project, the museum conservators informed us that the map was very sensitive to light exposure. Moreover, the map still had the original pencil markings of the first Duke of Wellington on it (as well as blood from one of the soldiers who was carrying the map on his body when he was killed in battle), so all measures should be taken to avoid damage to these materials.

One measure we took was to control the light exposure, of course, but another was to control the relative humidity inside the frame, giving rise to two benefits:-

1. This minimises any mould or insects from damaging the map, thanks to the controlled microclimate inside the display frame.
2. The second issue is due to the inherent ability of glass to collect electrostatic charge, which builds up when we walk across a woollen carpet, for example. This is the 'triboelectric effect'.

This can result in very high electrostatic voltages accumulating on our bodies, reaching up to 30 kV. Ensuring the humidity remains between a target of 45%-50% stops the air from drying out. If the humidity were to fall below that figure, the drier air, containing less water, can prevent any electrostatic charge on highly insulative surfaces (like acrylic and glass) from dissipating naturally to electrical earth through the water molecules in the air.

In the worst case, the resulting electric field can ‘lift off’ friable media on artworks, like chalk, charcoal and the graphite pencil markings made by the first Duke of Wellington on the Waterloo map.

In this case, since the smart glass is in very close proximity to a 200 year old original map of iconic importance to British military history, we took great precautions to ensure the smart glass did not damage the fragile pencil markings due to any electrostatic fields on the glass interior surface.

[edit] What does the future hold for smart glass?

Our collective journey into adaptive, intelligent smart glass building facades, smart glass showcases and smart glass enabled transportation has just begun. We have merely scratched the surface of the available use cases so far. Continued advances in materials and algorithms will benefit sectors where smart glass is already being used; from transportation to retail; from hospitality to heritage to healthcare.

What is really exciting is that global initiatives such as the push towards sustainability and user privacy are driving us to a future that may be beautifully illuminated by natural daylight, managed judiciously by smart glass.

[edit] Author Bio

Manoj’s first experience with glass goes back to the fabrication of optical waveguides as a final year project at the Optoelectronics Research Centre (Southampton University) in 1989, sponsored by Ferranti Semiconductors in the UK.

Manoj worked initially as a Trainee Patent Agent in London representing semiconductor firms at the UK, European and US patent offices, before embarking on a career as a Software Engineer across telecommunications, automotive, finance and eGovernment sectors.

Manoj founded ArtRatio in 2008 to build smart glass display cases for museums and luxury retailers, and launched SmartGlassWorld thereafter to promote development in smart materials.

Manoj’s experience as a manufacturer of smart glass end-products and as a smart glass distributor & consultant allows for a deep understanding of what customers need in order to achieve a return on investment in smart glass technologies.

Manoj is a UK Chartered Engineer with a Bachelors in Electronics Engineering from Southampton University and a Masters in Software Engineering from Oxford University, and is a guest lecturer on the MA Art Logistics programme at Sotheby's Institute of Art in London.

--Smartglassworld 15:27, 05 Sep 2022 (BST)