Since the very first immersive online platforms appeared in the 2000s, the metaverse has undergone several waves of development. These virtual worlds were initially designed for entertainment or social interaction, but there is now increasing interest from the industrial sector. These worlds can reproduce and synchronize physical environments in real time, in a virtual world.
This is crucial because it will improve decision-making thanks to accurate simulations, it will accelerate innovation and enhance collaboration by making complex data more accessible and visually available. As opposed to traditional industrial digital twins, the evolving metaverse models can detect and integrate any events (changes on a machine, a person passing by, etc.) to then guide operational decisions. Five Institut Mines-Télécom schools, several Airbus subsidiaries and Orange are involved in the 5GMetaverse project, with the aim of preparing and promoting 5G for the needs of the metaverse, specifically in an industrial context.
A dynamic version of the industrial digital twins
The metaverse can offer industry a whole range of possibilities, including simulation and optimization of processes before they are rolled out in the real world. “There is a great deal of interest in simulating the various settings of a machine to optimize its operation in a controlled and controllable virtual environment”, says Marius Preda, researcher specialized in augmented reality at Télécom SudParis and involved in the 5GMetaverse project.
Remote operation and remote assistance are also promising applications and are included in the use cases for the 5GMetaverse project. An operator equipped with Augmented Reality (AR) glasses can, for example, be guided remotely by an expert thanks to a virtual representation of the industrial environment. “The expert works in a digital twin of the installation in real time, and their interactions with the virtual world are communicated to the operator on-site”, explains Marius Preda.
In addition to improvements in efficiency, this technology also enhances human-machine collaboration. As industry is evolving and becoming increasingly robotized, the metaverse provides an interface that facilitates interaction between humans and robots, while exploiting human adaptability in environments where robots remain limited. This can only be achieved if the metaverse overcomes major technological obstacles, and manages to meet the increasingly demanding technical requirements in terms of latency and throughput.
Latency and throughput: key reactivity indicators
Considering that every action that takes place in the real world needs to be instantly transferred to all users of the metaverse in real time, requirements in terms of latency and throughput are particularly crucial. “In a Zoom meeting, each user requires around 1 Mbps for a good quality video experience. But to produce a realistic image in a metaverse, it’s necessary to communicate every physical movement and this requires at least 20 Mbps per user”, specifies Marius Preda. “Latency also needs to be very low to ensure a seamless experience. If my body moves even half a second after I have spoken, the difference is highly visible.”
These needs obviously vary if we’re talking about virtual reality (VR) or AR. VR plunges the user into a completely virtual environment, whereas AR overlays virtual elements onto the real world. VR therefore requires a lot of content and fast throughputs because it recreates a fully immersive environment. The requirements for AR are less demanding but this still requires immediate synchronization with reality. In the 5GMetaverse project, VR is used to provide experts with a fully immersive experience, and AR is used for operators in the field, meaning the requirements in terms of latency and throughput differ.
Optimizing the flow of and prioritizing critical data
The other challenge concerns data management. Identifying priority information is just as important as guaranteeing fast data transfer. “In an industrial environment, certain data such as machine control are more important than others”, specifies Marius Preda. “Transmission protocols must also consider data priority in order to optimize data flows.”
But there again, the requirements differ depending on the use. For example, the needs of an operator interacting remotely with a machine are different to those of a training session, because a higher latency can be tolerated. Likewise, some machines require high levels of interaction, whereas others are more independent. It is therefore necessary for industry experts to define critical data so that the metaverse knows to give this data priority, which therefore optimizes data transfer in accordance with the importance of each dataset.
Key technology for a metaverse that can meet industrial needs
To meet these needs, several types of technology are required. Data compression technology plays a key role as it reduces the size of data and speeds up the transmission process. 5G allows for better data flow management thanks to network slicing, which allows data flows to be divided into multiple ‘slices’ tailored to specific needs. However, this technology is limited, and it will need to evolve to 6G for example, if it is to remain suitable for the wide range of uses and be rolled out on a broader scale.
These changes should also facilitate interoperability between different metaverses. In the case of remote assistance for example, the expert will need to pass from one metaverse to another to be able to provide assistance across different locations, meaning the management of their profile and login details needs to be seamless. The changes will also need to prepare for the integration of multiple users into the interface design, which is key for environments such as the virtual factory which hosts a dozen users at once, or the virtual shop – second use case of the 5GMetaverse project – which could welcome thousands.
Finally, artificial intelligence (AI) is also key in harnessing the full potential of the industrial metaverse, and it is used at several levels, including the classification of data to optimize data processing or the representation of 3D content. “AI can for example generate, using one single image, a 3D representation of a machine or a component, therefore taking the pressure off requirements in terms of throughput since certain data is no longer transferred, it is generated”, illustrates Marius Preda.
Working towards the ‘fifth wave’
The industrial metaverse is certainly looking promising for the future, but its widespread deployment is a goal for the medium or long term. “The overarching aim of the metaverse is to provide a fully immersive experience in a virtual world, not to create an upgrade of Zoom”, says Marius Preda. “To do so, the interfaces provided by virtual reality headsets are key”. The current devices are still too bulky and not really adapted to extended use in an industrial context.
The future of the metaverse in industry therefore depends on progress in several fields all at once. In terms of hardware, the size of equipment needs to be reduced and the equipment adopted on a widespread scale. In terms of networks, new protocols have to be integrated, with 6G for example, and finally, progress has to be made in AI and modeling to enrich the available applications. According to the researcher, we will have to be patient and wait for the ‘fifth wave’ of the metaverse [see below]: “not for at least another 7 to 10 years”.
