Recently, you may have come across the term “quantum computer” more frequently. Quantum computers are a next-generation technology still in the research and development phase, but they are expected to perform high-speed computations, drawing significant attention.

Quantum computers can be broadly categorized into two types: “gate-based” and “annealing-based^*1 “. The “gate-based” type is designed to handle general computational tasks and is expected to function similarly to modern classical computers.

On the other hand, the “annealing-based” type is specialized and excels at solving the “combinatorial optimization problems”, which are prevalent across various domains. Due to its ability to solve these problems efficiently, annealing-based quantum computing is gaining attention in fields such as logistics, transportation, drug discovery, and finance.

*1 The term “annealing” comes from the process of metal annealing, where metal is heated and then gradually cooled to relieve internal stress and achieve a stable state through recrystallization. This concept of “thermal fluctuation” is similar to the process of controlling “quantum fluctuation” seen in quantum mechanics over time to gradually weaken the fluctuation and search for the optimal (lowest energy) state, hence the term “quantum annealing.”

Currently, the commercial adoption of annealing-based quantum computers (hereinafter referred to as quantum annealing computers) is becoming a reality. At the same time, the practical application of Simulated Quantum Annealing (SQA) computers— which simulate the computational algorithms of quantum annealing computers on classical hardware—is also emerging.

Sharp has a department dedicated to developing the AGV*² Operating System (AOS) using SQA machines. To learn more, I interviewed Kurimoto and Nguyen from Engineering Division I, Robotics Business Development Unit, Smart Business Solutions BU, who are responsible for SQA machine development.

*2 Automatic Guided Vehicles (AGV) that carry goods and parts in warehouses, also called transport robots.

■Why are you developing the SQA machine?

(Kurimoto) With the rapid rise of the e-commerce (EC) market worldwide, the logistics industry is facing an increase in the number of deliveries, leading to the need for large-scale warehouses and automation to address labor shortages. As a result, there is a growing demand for large-scale control of AGV used in warehouses.

Our current optimization solution for sites with 500 AGVs meets the current needs, but we believe the time will soon come when this will no longer be sufficient.

The development of quantum annealing computers, which are promising for solving this issue, is progressing. However, the number of quantum bits^*3 in commercially available and the most advanced quantum annealing computers is about 5,000, which can only handle the route calculation for about 100 AGVs. To solve the route calculation for more than 1,000 AGVs as an optimization problem, tens of thousands of quantum bits are required.

*3 A bit is the basic unit (one digit) of binary notation, which represents numbers using 0 and 1. For example, two digits (00 or 11) are 2 bits, and three digits (000 or 101) are 3 bits. A quantum bit is the basic unit representing numbers in a quantum computer.

Sharp is developing the SQA machine to meet the needs of controlling more than 1,000 AGVs by performing high-speed calculations for combinatorial optimization problems.

The introduction cost of a quantum annealing computer capable of calculating for more than 1,000 AGVs is expected to be in the billions of Japanese yen, making it difficult to invest in the logistics field. Therefore, we consider the SQA machine to be more realistic.

■What is a quantum computer?

(Nguyen) The SQA machine simulates the calculation algorithms of quantum annealing computers on classical computers, so let me briefly explain quantum computers and quantum annealing computers.

Quantum computers use the unique physical states of quantum mechanics observed in the microscopic world of atoms and electrons to achieve high-speed calculations, utilizing superposition and quantum entanglement states.

Specifically, they use special bits called quantum bits for calculations. While classical computers use bits in the state of 0 or 1, quantum bits can create a superposition state where 0 and 1 exist simultaneously, representing many states. By leveraging this property, quantum computers can process multiple calculations simultaneously, which is their greatest feature.

■What is a quantum annealing computer?

(Nguyen) Quantum annealing computers are specialized computers that solve optimization problems by applying the principles of quantum mechanics. Specifically, they model problems as energy states and search for the lowest energy state. Quantum bits can overcome energy barriers that classical approaches cannot, potentially reaching better solutions through quantum tunneling effects^*4.

However, in order to perform computations on a quantum annealing computer, it is necessary to model various problems for quantum computation and to develop and implement quantum algorithms tailored to these models.

*4 Reference: Experimental Evaluation of Path-integral Quantum Monte Carlo for Restricted Boltzman Machine （Japanese）

■Tell us about the modeling of AGV route calculations and the development and implementation of quantum algorithms.

(Nguyen) The route calculation for multiple AGVs can be modeled and reduced to a combinatorial optimization problem, to which quantum annealing calculation methods can be applied.

In our AGV Control System (AOS), we use the widely known Dijkstra’s algorithm for shortest path calculation, which is commonly used in car navigation systems to calculate the shortest path by assigning weights to routes.

Currently, for route calculation of multiple AGVs, as shown in the figure below (a), we sequentially calculate the routes for AGV #1, AGV #2, and so on, and issue travel instructions to each AGV. In this method, travel instructions are issued for each AGV’s route calculation, and for example, if there are 1,000 AGVs, the next one cannot proceed until the 1,000th calculation is completed.

Therefore, to reduce the AGV route calculation to a combinatorial optimization problem, we replace it with an Ising model^*5 and set the objective function.

*5 A mathematical model used in statistical mechanics to understand phase transitions in magnetic systems, which can also be applied to model combinatorial optimization problems.

By solving this calculation using the quantum annealing method, we can simultaneously calculate the routes of AGV (appropriate route combination calculation) as shown in the figure above (b), achieving speed and overall optimization.

This is an image of parallel computation, considering all route calculation candidates simultaneously and deriving the appropriate “route combination” from them.

Here, there is a problem. Although it is theoretically possible to speed up using quantum annealing, as of 2025, the development of hardware to realize the calculation has not caught up.

As mentioned, the current number of quantum bits in quantum annealing computers is approximately 5,000. To solve the optimization problem of route calculation for 1,000 to 2,000 AGVs, it is anticipated that the required number of quantum bits would be in the range of several hundred thousand to one million. Therefore, the current number of quantum bits is insufficient for application.

■Can the SQA machine solve this problem?

(Nguyen) Yes. Sharp is developing an SQA machine that implements quantum algorithms simulating the optimal solution search method of quantum annealing on classical computers, which currently cannot perform large-scale route calculations.

This SQA machine has a calculation speed thousands of times faster than “general-purpose classical computers.” Although its absolute calculation speed does not reach that of quantum annealing computers, it can reproduce simulated quantum bits exceeding the current number of quantum bits in quantum annealing computers, making it possible to handle large-scale problems that current quantum computers cannot.

This development is being advanced through joint development with Tohoku University, one of Japan’s leading quantum computer development centers^*6.

*6 Please refer to the news release.

   https://corporate.jp.sharp/news/231219-a.html　（Japanese）

Specifically, we aim to realize an SQA machine that can execute quantum algorithms by developing quantum algorithms (including modeling optimization problems) that leverage the unique properties of quantum bits on classical computers.

The SQA machine under development uses electronic circuits (FPGA circuits^*7) capable of performing parallel calculations at high speed and implements quantum algorithms in a special circuit design, enabling the simulation of quantum annealing on classical computers.

*7 Field-Programmable Gate Array. A computing device (integrated circuit) that can be reconfigured by the user to create custom circuit configurations.

■What are your future prospects?

(Kurimoto) The SQA technology we are working on is not limited to AGV route calculations in current logistics warehouses but can be applied to a wider range of areas.

For example, in logistics warehouses, the arrangement of stored products and the order of inbound and outbound operations significantly impact overall efficiency. In large warehouses handling tens of thousands or millions of products, combinatorial optimization considerations become very important, but classical computer calculations cannot keep up.

In the future, I believe the effectiveness of the SQA technology introduced will be demonstrated in addressing various increasingly complex social issues. Please look forward to it.

— Thank you very much.

In the future, when quantum computers are completed, I had the impression from the news that calculation speeds would dramatically increase, but I had no understanding of the current situation until this interview.

In this interview, I learned that the SQA machine for optimizing AGV control in logistics warehouses is almost ready for practical use.

I also understood that combinatorial optimization problems exist in various fields and have a wide range of applications. I sincerely hope that these technologies will be put to practical use as soon as possible.
(Public Relations C)

Product Information

 Automatic Guided Vehicles (AGV) (Japanese)