Hundreds of Quantum Applications

Abhishek Awasthi∗, Nico Kraus†, Florian Krellner‡, David Zambrano†

*BASF Digital Solutions GmbH, Ludwigshafen am Rhein, Germany

†Aqarios GmbH, Munich, Germany

‡SAP SE, Walldorf, Germany

This work is a benchmark study for quantum- classical computing method with a real-world optimization problem from industry. The problem involves scheduling and balancing jobs on different machines, with a non-linear objective function. We first present the motivation and the problem description, along with different modeling techniques for classical and quantum computing. The modeling for classical solvers has been done as a mixed-integer convex program, while for the quantum-classical solver we model the problem as a binary quadratic program, which is best suited to the D-Wave Leap’s Hybrid Solver. This ensures that all the solvers we use are fetched with dedicated and most suitable model(s). Henceforth, we carry out benchmarking and comparisons between classical and quantum-classical methods, on problem sizes ranging till approximately 150, 000 variables. We utilize an industry grade classical solver and compare its results with D-Wave Leap’s Hybrid Solver. The results we obtain from D-Wave are highly competitive and sometimes offer speedups, compared to the classical solver.

Inadequately planned factory layouts can lead to higher operational costs. The planning process is time-consuming and multiple planning parameters must be considered simultaneously, leading to large solution spaces. Automated planning approaches can support the planning team in the early phases of layout planning by generating layout variants that can afterwards be further specified. Recent studies from other disciplines have shown the potential of Quantum Annealing (QA) to solve complex assignment problems within seconds. Consequently, this paper presents a first implementation of a QA-based layout planning approach and demonstrates its advantages regarding scalability, solution quality, and computing time.

Keeping GE’s robust R&D program running presents significant logistical challenges, especially when it comes to asset sustainment. Using D-Wave’s quantum computing system, the company set out to improve resource allocations for machine repairs and minimize resource contention when multiple repairs are requested at once.

As part of their work supporting Innovate UK, British multinational defense, security, and aerospace company BAE Systems undertook an assessment of how to optimize planning and scheduling of operations, distribution, and telecommunications across dozens of major industries using quantum computing.

When BMW began looking for ways to optimize their production lines, they turned to D-Wave. Using quantum algorithms, the company streamlined the complexity and requirements for programming robots used in PVC application and joint sealing.

In addition to creating smarter, safer, more environmentally friendly automotive components, DENSO envisions incorporating quantum computing into everything from production and robotics to scheduling and logistics for product delivery. Work is already underway to make that vision a reality. DENSO recently completed proof-of-concept work using D-Wave's quantum platform, aimed at optimizing control of Automated Guided Vehicles (AGVs) on their factory floors.

In 2019, DENSO and Tohoku University announced the creation of an algorithm designed to improve the ability of a D-Wave system to handle complicated problems. The algorithm partitions an original, large problem into a group of subproblems. The D-Wave system then iteratively optimizes each subproblem to eventually solve the original, larger one. DENSO plans to use the algorithm to address everything from public transportation to the coordination of robotic systems in factories.

Given a set of jobs and a finite number of machines, this demo show how to schedule jobs on those machines such that all the jobs are completed at the earliest possible time.

On the Volkswagen factory floor, the production process creates a series of constantly changing conditions. Delays, shifting wait times, and irregular requests all impact business objectives and outcomes. In order to find optimization solutions, Volkswagen is developing a hybrid algorithm that relies on both classical and quantum approaches to problem solving.

This is a heuristic approach on how to optimally schedule jobs using a quantum computer. Given a set of jobs and a finite number of machines, how should we schedule the jobs on those machines such that all the jobs are completed at the earliest possible time?

Toyota CRDL developed a hybrid quantum-classical algorithm for model predictive control (MPC). Unlike the usual application scenario for MPC algorithms the research team used the D-Wave system for dynamic control problems in which solutions must be obtained at high speed, specifically stabilization of a spring-mass-damper system, such as the ones in our cars that absorb shock on uneven roads; and dynamic audio quantization for radios and GPS devices. In both cases, the D-Wave method exhibited better performance than two classical approaches.

Previous work by GE Research has looked at using quantum computing to do resource and path planning to improve the efficiency of our asset sustainment operations around things like jet engines,  wind turbines and medical equipment repairs.  The work was meant to be exploratory and our projections showed it taking a few more years before it would be competitive with other approaches.  With access to the latest generation of hardware and hybrid algorithms we are beginning to believe that our projections are too conservative and that the time for industrial use of quantum computing has already started.