What could quantum computing do for Automotive Companies?
In the final post of this series on the rise of quantum computing, Bahram Ganjipour, a senior software engineer and quantum computing lead at Volvo Cars, goes over some of the real-world use cases of this fast-moving technology — as well as the current limitations on its potential.
What exactly can quantum computing assist us with?
As highlighted in the first post, one of the main hurdles for any organization when it comes to quantum computing (QC) is the need to comprehend the business potential of quantum computing. This understanding is crucial for making informed decisions regarding the adoption of quantum computing technologies. Early adoption of quantum computing can help companies in navigating this complex landscape and distinguishing between hype and real opportunities. By embracing QC early on, organizations can gain a competitive edge and position themselves at the forefront of quantum advancements.
Real-world use cases of quantum computing and Potential game-changer:
When it comes to tackling real-world business problems using quantum computing we need to consider several elements listed below:
- How to map a business use case into a format that can be handled on quantum hardware. Constrained combinatorial optimization problems may be encoded in a constraint-free form.
- Being aware of and selecting quantum algorithms compatible with NISQ hardware.
- Developing programming skills for quantum devices. Both direct proficiency in a certain language and conceptual understanding of how to think when developing quantum algorithms.
- Simulation/execution, and what you can learn from it, such as how quantum performance compares to classical performance and the potential for mid-term, non-trivial achievements.
o Machine Learning and large datasets
The concept should be of particular interest to anyone who works with data science at scale. Machine learning (ML) can be immensely valuable in analyzing massive amounts of data as handling the complexity of analyzing large data sets can be overwhelming for humans. ML algorithms require a sizable amount of historical data as well as need to be fed with information to learn and recognize trends over time. However, as the volume of data grows, the complexity of the computation also increases, leading to increased processing times and challenges in identifying, interpreting, and presenting meaningful results. The limitation of ML algorithms is primarily due to the computing power of conventional computers. However, quantum computers are capable of processing large data sets at a considerably faster speed, providing AI systems with greater granular insight for identifying patterns and anomalies in data analysis.
Machine learning (ML) algorithms that are inspired by quantum computing use the ideas of quantum physics to develop new, more effective algorithms for processing data. The benefits of quantum-inspired algorithms in ML include enhanced optimization using quantum annealing and variational quantum algorithms, better feature mapping with methods like quantum feature mapping, improved classification with quantum support vector machines, increased privacy using homomorphic encryption, the ability to construct more potent and effective neural networks with quantum neural networks, and more efficient data processing with algorithms inspired by quantum mechanics.
o Optimization problems
One class of mathematical problems is optimization. Optimization problems are the most natural kind of problems that you might think of at first when you think of quantum in an automotive context. They are the basis of routing, logistics and supply chain, and manufacturing (such as PVC sealing, Paint shop…) optimization etc.…
There are tons of problems in which at best we do approximations, but we are far from optimal because the number of possibilities is enormous.
An optimization use case is robot path planning: A lot of the production steps are handled by robots such as PVC sealing and paint shop. Robot path planning is an optimization challenge that can be solved to accelerate production lines, leading to higher throughput without the need for additional resources. This development in automation technology has the potential to greatly enhance manufacturing operations and productivity, providing significant benefits to the industry.
One example of robot path planning is PVC sealing: PVC sealing is a process where a plastic material is applied to seal the seams formed by metal sheets in automobiles, thereby preventing water from entering the vehicle. Typically, robots are employed to carry out this task, and robots are programmed to follow specific paths along the vehicle body. The challenge is to find collision-free paths for multiple robots while optimizing time efficiency to speed up the production process and generate significant business benefits.
The complexity of this problem arises from a number of reasons. Each workstation may have 3 to 6 robots, each with different nozzles for different types of PVC. Moreover, there can be up to 50 different seams per region that need to be sealed, and different car models may have varying seam configurations. As a result, a large solution space needs to be addressed, requiring different plans for each robot and even different configurations of the same vehicle type. Currently, heuristics and nearest-neighbor approaches are used to solve these problems. However, recent research has shown that the lower bound of achievable solutions is much lower than the current practices. Today’s solutions are often constrained by computing power, and consequentially limited in size and scope. These optimization tasks also happen to be the kind of challenges that today’s quantum computers might turn out to be game-changers.
o EV Battery
Simulating the behavior of molecules and materials at the quantum level, a critical aspect of battery chemistry becomes increasingly challenging as the system size grows when using classical methods. However, quantum computers leverage inherent quantum properties to perform these calculations with remarkable efficiency. This enables them to accurately model interactions among all the particles within a battery’s components, ultimately leading to precise predictions of electron and ion behaviors within the battery. Consequently, this capability facilitates superior battery optimization and yields chemistry insights that were previously beyond reach through classical approaches.
So why “could” and not “is”? Part of the reason is that commercial quantum computing is still in its infancy. Quantum computers are not widely available and those that are accessible via cloud come with limitations. The limitations are the finite number of qubits, and noise which can cause errors in quantum calculations, hence limiting the processing power of quantum computers. Beyond the technical challenges, the industry is still in the process of defining how to program quantum computers. This concerns not only the programming language but also our ability to identify and frame real-world business use cases in a way that can be tackled by quantum computers.
This is only the state as it is now. The rapid pace of technological development in the last decade has taught us the landscape can change quickly. Particularly when tech giants such as Amazon, Microsoft, and Google are involved. We might not be there, but we are moving into an era in which companies across various industries can use quantum computing to create a major competitive edge.
If quantum computing resources are not readily available or feasible to implement in the current context, can quantum-inspired algorithms (QIAs) serve as an interim solution?
The answer to some extent is ‘’YES”. Quantum-inspired algorithms can serve as a viable alternative to some extent. Quantum-inspired algorithms, although not providing the full exponential speedup of actual quantum computers, enhance the efficiency of classical algorithms by drawing inspiration from quantum principles and techniques.
What are QIAs? What is their relation to quantum computing?
Quantum-inspired algorithms are classical algorithms that rely on linear algebra techniques, particularly tensor networks, which have been developed in recent years, thus making the classical machine operate algorithms that benefit from the laws of quantum mechanics that benefit real quantum computers. If you want to know more about tensor networks, see this site.
These algorithms enable businesses to harness the advantages of quantum computing while running on classical computers. They mimic certain aspects of quantum computation and leverage the power of tensor networks, which provide a structured and efficient representation of high-dimensional data, making them well-suited for optimization tasks.
By combining the concepts and techniques from tensor networks with classical optimization algorithms, quantum-inspired algorithms can achieve improved efficiency. The structured representation provided by tensor networks allows for a more compact and efficient representation of high-dimensional data. Instead of directly manipulating and processing each individual data point, tensor networks capture the correlations and relationships among data points, reducing the computational complexity involved.
The ability of quantum-inspired algorithms to handle high-dimensional data more efficiently than classical approaches opens up new possibilities for automotive companies. It allows for advanced data analysis techniques, such as pattern recognition, anomaly detection, and predictive modeling, to be applied to large and complex datasets. This, in turn, can drive innovation, improve decision-making processes, and unlock hidden insights that may not be readily apparent through conventional computational approaches.
Through this synergy, quantum-inspired algorithms can push the boundaries of optimization capabilities, offering a promising avenue for tackling complex problems in diverse fields including combinatorial optimization problems in the automotive industry.
So, we at Volvo Cars should ask ourselves: what do we see as our journey through quantum? If the focus is on harnessing the latest advancements in computational problem solving, possibly driven by quantum computing, to tackle the organization’s present challenges, then advocating for the adoption of quantum-inspired solutions can be a reasonable approach.
However, if the goal is to truly explore the boundaries of classical computation within the business and prepare for the potential disruption of quantum technology, then settling for quantum-inspired approaches would not suffice.
Conclusion:
It’s thrilling to embrace quantum computing at this time. In terms of hardware, methods, and the capacity to incorporate quantum computing’s capabilities into practical applications, there will probably be significant advancements in the next years, some anticipated and some unforeseen. Nevertheless, the promise of a paradigm shift in computation requires a change in our perspective and skill set to lower the adaptation barrier.
I appreciate you taking the time to read these posts, and I wish you fun as you go along the quantum learning curve.
