Liu Jianya and Guo Liang: “The carbon footprint of the metaverse can be reduced”

Interview by Chen Xiaorong

UNESCO

If we set an analogy, what is the relationship between the metaverse and mathematics?

It's been nearly one year since Facebook announced it was rebranding to Meta and would focus its future on the upcoming ‘metaverse’. Since then, what that term means hasn't gotten any clearer. 

Mathematically, we define that the metaverse is a function that is represented by two sets of objects with arrows drawn between them to show the relationships between the objects: one set represents objects in real life, say, the palace of Versailles. The second one refers to computer models of real-world objects, such as a digital palace of Versailles, which can be shown and manipulated on the screen.

In short, metaverses can be regarded as the mathematical operation that associates each element in the real world with one or more elements in the digital or virtual world.

Can we fully realize the metaverse with the current computing power?

Maybe, but for what purpose and at what cost? Broadly speaking, there are two technologies related to the metaverse: data acquisition and virtual reality technologies. Data acquisition is the technique of capturing the shape and look of natural things in computer vision using cameras or laser scanners. Virtual reality (or digital twin) means generating digital objects to reconstruct the physical worlds. Both techniques are computationally expensive. Both heavily rely on artificial intelligence models and cloud services that require quite large amounts of energy. 

A recent study conducted by researchers at University of Massachusetts estimates that training just one AI model could generate 284 tons of carbon dioxide, which is more than five times the amount of greenhouse gases emitted by a car in its lifetime. Cloud computing, which is necessary for virtual reality, online gaming and high-resolution image processing, could also significantly raise carbon emissions.

Therefore, the metaverse should be held environmentally accountable. A 3D reconstruction of the palace of Versailles is useful, as a virtual tour website enables people from all over the world to plunge into interactive frescoes and to discover the paintings, sculptures, and engravings in a new way. On the contrary, there is no point in wasting energy on a digital twin of an ugly town hall, as there is no need for a citizen to wear a virtual reality gadget to ‘walk’ in the digital mock-up of a dreary concrete building to access public services. 

What is the environmental impact of the metaverse? 

The metaverse is one of today's hottest technological and socioeconomic topics. Lots of companies are already working on creating services for this new digital world. However, the applications of metaverse-related technologies like artificial intelligence (AI), virtual reality (VR), 3D animation, blockchain and many others, are still human-centric: people make decisions that prioritize humanity over the environment. 

Furthermore, artificial intelligence and its supporting systems cause increased environmental costs. It is becoming increasingly energy and computation intensive to train deep learning models that use artificial neural networks to work with large datasets. Thus, financial and environmental concerns are growing. 

As the metaverse becomes more complex, we must use even more data. There’s a problem here, since data centers use an incredible amount of energy. It's unclear how much energy is required to store the data generated for and by the metaverse, but the number is likely staggering. A data center’s building and cooling systems also produce a lot of CO2.

In short, metaverse is energy-intensive, and the higher the demand for it and its related technologies, the more power we use. It is the responsibility of the technology industry and researchers to learn from the environmental impact of metaverse. In all technological decisions, we must factor in the earth experience. 

Can mathematics contribute to a more environment-friendly metaverse?

Mathematics can be used in various ways to reduce the energy consumption of the metaverse. For instance, the method from Nanyang Technology University researchers in Singapore focuses on selective scanning to create virtual environments. Instead of transmitting the entire captured image, the camera first automatically selects the objects of interest and transmits only these objects to metaverse service providers. For example, to transmit images or data of a public transportation scene, pedestrians and vehicles would be selectively scanned by innovative calculation, whereas other objects in the scene would require less calculation and energy. 

Our team at Shandong University has been working on a sampling method derived from analytic number theory to reduce the power consumption of metaverse technologies. Our focus is on laser scanning, which is the most effective way to create digital representation and 3D digital models for metaverse.

A laser scanner emits a beam of infrared laser light onto a rotating mirror that paints the surrounding environment with light. Objects in the path of the laser reflect the beam back to the scanner, providing the geometry that is interpreted into 3D data. At the same time, the scanner head rotates and sweeps the laser across the object’s surface, creating a multitude of points. It is computationally expensive to register, display and process these large point clouds.

Let's say we intend to generate the metaverse of the obelisk in the Place de la Concorde in Paris. To create the digital twin, a laser scanner would normally need to produce one million measurement points to obtain the precise parameters of the obelisk. Using our method, the scanner can produce 40 per cent less measurement points while maintaining the same accuracy of a digital twin produced using traditional methods. This method allows us to significantly reduce the energy and time needed to create a metaverse. Faster computing means less carbon emissions.