Skip to Content

A Quantum of Intelligent Industry

Mike Dwyer
17 Mar 2022

In the first of our “Intelligent Industry: Journey to Farnborough International Airshow” blog series, Mike Dwyer considers the potential impact that the world of quantum computing, sensing and communication could have on our ability to create new intelligent products and services.

The world of quantum is complex, strange, and counter-intuitive. The late eminent physicist Richard Feynman said, “If you think you understand quantum mechanics, you don’t understand quantum mechanics”, therefore this is not a blog to outline a shortcut to understanding the intricacies of Feynman diagrams and energy states! Instead, let’s imagine a world where quantum computing, communication and sensing supports the skilled people that design, build, test, service and operate highly complex systems, or sensitive machinery, or manufacturing processes. On a daily basis, how can we unlock the person’s creativity, decision making, and assurance skills?

Capgemini’s Intelligent Industry proposition is the evolution of Industry 4.0 with human-centricity and sustainability becoming the core focus for immersive and intuitive industrial systems and ways of working that are powered by trusted data. To deliver connected products, supply chain, operations, and services more effectively, faster, and at a better price/cost point, we need a quantum leap in our thinking and approach to both the large scale and small scale “knotty” engineering problems. Let’s think about digital twins, secure communications, sensing and machine learning…

Quantum Computing for large-scale Digital Twins

What If I: operate a national scale distribution infrastructure, how do I make true real-time decisions, how do I factor in global events, weather, or emergent trends such as electric vehicles, changing work habits, to ensure stability of supply? How do I optimise this?

Digital Twins are a core component of Intelligent Industry; they connect the theoretical and real worlds together and allow us to make sense of complex data, patterns, decisions and ‘what if’ scenarios – for example for a product, fleet, network, or organisation. We foresee a Digital Twin or series of Digital Twins that would be able to model the network, ingest the real-time demand/supply data, overlay other dynamic data, and then make immersive and accurate decisions.

Building a very large-scale Digital Twin of a distribution network is a huge but not insurmountable challenge. The Digital Twin will need many elements to come together including organisations, processes, assets and real-time data, data standards, systems integration, as well as end-to-end Cybersecurity from asset to Digital Twin. The larger, more complex the network model, the greater the computing power needed to “solve the equations” in, ideally, real-time. This creates a tipping point between using hyperscale or high-performance compute capability vs a quantum computer.

Quantum computing is a fundamentally different concept to large scale computing that naturally facilities multi-dimensional complex problem solving. Future use cases could include everything from complex optimisation and simulation to new era of quantum AI, with applications spanning everything from aircraft design to drug discovery.

A Quantum Digital Twin could be used for network optimisation for energy provision, network scheduling via integration with; weather models (hot\cold\stormy etc.), global live events on TV, renewable energy sources to support in grid stability and, decentralised prosumers (consumes and produces) feeding into it. The national and international focus on global climate change and sustainability could also be modelled to understand the tangible impacts on grid resilience and readiness.

Capgemini Perspective: Digital Twins

Quantum Communications for secured comms

What If I: want to protect critical infrastructure such as sensitive networks or asset communications? What if I want to connect a very large number of assets and facilities together, how can we safely do this?

End-to-end cybersecurity in Intelligent Industry is a foundational tenet. We cannot exploit great insight, processes, and technology if we cannot trust the veracity and accuracy of the underlying data. This must be mitigated or eliminated, particularly if we are communicating across the public domain. The threats are evolving rapidly, and we therefore need to be at the forefront of this to ensure we can share information and, for instance, protect the integrity of long-lived measurement or safety devices in the network.

Using quantum communications means to transmit and control information through the laws of quantum mechanics such as QKD (Quantum Key Distribution). This is the only cryptographic primitive so far proven secure that works at the hardware layer. Another approach could be Post-Quantum Cryptography (PQC), which provides security at the software layer. Although commercial applications are available for both QKD and PQC, it’s still an emerging field and needs to be carefully researched and tested to ensure the unbreakable edge remains unbreakable.

Based on post-quantum cryptography and quantum key distribution, new security primitives can be developed for confidential data sharing and computing. One such example is a project at Capgemini on Quantum Distribution Oblivious Transfer (Q.DOT). Quantum distributed oblivious transfer is a primitive for secure multiparty computation and enables a user to process data without having access to the underlying data (which may contain private information).

Thinking further into the future, quantum entangled communication will give rise to a quantum internet. Here, we’re very much looking at a similar situation to our existing internet in the 1980s, as we’re still exploring how widespread sharing of entanglement could be used. However, potential applications could include distributed quantum computing, remote sensing networks or quantum secured money. This would be a quantum leap in Intelligent Industry cybersecurity!

What’s next for data, blockchain and quantum computing?

Quantum Sensing for extreme sensitivity

What If I: want to measure at a much higher sensitivity and smaller scale but on an industrial basis?

Quantum sensing is used to measure the disturbances at an extremely sensitive scale, which may reveal information that cannot be gathered today. This new view of the world will help drive new product and service innovations for areas such as medical diagnosis, drug discovery, material inspection and battery cell testing. Autonomous transport, facilities and assets will benefit from having a high-fidelity picture that goes beyond our senses, for example, seeing gravity and electric/magnetic fields. Measuring at quantum levels unlocks further benefits of miniaturisation, usability, and precision.

Over time, as these devices are industrialised, the cost-point will reduce and the impact on intelligent products and services will exponentially increase as we collectively unlock new applications and avenues. We could envisage a world where robots, automated vehicles and drones can survey complex facilities and assets and see into the material structure of all the components, not just the paint and surface corrosion. This would unlock remote inspections’ true potential to find and monitor fatigue, wear and tear and help make predictive / preventative maintenance even more powerful.

Quantum Machine Learning for engineers

What if I: want to push the limits of advanced traditional machine learning and achieve improved learning without massive data volumes?

Classical machine learning has enjoyed tremendous success over the last decade and is increasingly essential for a wide range of Intelligent Industry operations. However, classical machine learning has its limitations. Deep learning approaches such as computer vision and natural language processing require vast amounts of data. Large natural language processing models such as GPT-3 include 175 billion parameters and require the energy consumption of a small city to be trained. This is unsustainable and impracticable as the amount of (high quality) data is also often not available.

Quantum (inspired) algorithms offer the potential to improve machine learning. In the near term, classical but quantum-inspired models could compress the size of deep learning models to reduce the number of parameters to be trained and improve energy efficiency. In particular, tensor-network based machine learning can compress deep learning models with a limited impact on accuracy. Although these algorithms are entirely classical and do not benefit from quantum phenomena, there could be a modest improvement over traditional methods. Additionally, advanced machine learning methods could be a bridge to quantum algorithms.

Machine learning algorithms on quantum hardware have also enjoyed significant academic and quantum community interest. Over the past few years, numerous new quantum machine learning (QML) algorithms have been proposed, including quantum support vector machines (qSVM), quantum natural language processing (qNLP) and quantum convolutional neural networks (qCNN). These algorithms can best be understood as kernel methods transforming classical data into (higher dimensional) quantum data. Specific problems might be better expressed in the quantum space, resulting in improved learning for QML with reduced data. Fortunately, many quantum machine learning algorithms might be applicable in the near term. That’s because these algorithms are hybrid, meaning that part of the workload is transferred from the quantum computer to the classical computer, so the demands on quantum hardware are reduced. Additionally, noise from imperfect quantum hardware might result in better generalisation of QML, whereas noise would result in errors and dysfunctioning conventional algorithms.

Machine learning on classical computers and eventually quantum computers is a promising area for the Intelligent Industry. Deep learning is already impactful for computer vision and natural language processing, we expect additional opportunities of (advanced / quantum) machine learning in all areas ranging from R&D, QA and operations. As such, we expect new machine learning to perform tasks previously impossible, as well as to improve the performance of conventional methods such as finite element methods and density functional theory.

Capgemini Perspective: Intelligent Industry

What is Capgemini doing with Quantum?

The convergence of the quantum world and Intelligent Industry is exciting. The pace of advancement means that practical applications of quantum-based computing, communications, sensing, and machine learning are unfolding, and moving from academia to business implementation. Capgemini’s Quantum Labs in Cambridge and Portugal are working on many similar use cases with partners such as IBM, where we are pushing the boundaries and developing new capabilities. If we are going to accelerate our development of products and services to meet the highly complex needs of our customer and consumers, we need to unlock a Quantum of Intelligent Industry.

More details can be found here:

Mike Dwyer

Head, Intelligent Industry CoEx UK

Mike is the head of the Centre of Expertise for Intelligent Industry in the UK and brings 25 years of industrial and consulting experience to his wide range of aerospace, defence, industrial, life sciences and energy clients. Mike started his career at Rolls-Royce plc before moving to Siemens PLM Germany and then into management consulting. Teating