Robotic Inspection
D-Wave Quantum Annealers Speed Up Industrial Optimization and Hold Promise for Difficult Problems
With businesses like D-Wave Systems expanding the frontiers of industrial applications with their hybrid quantum-classical algorithms and quantum annealers, quantum computing is developing quickly. The potential of these technologies to resolve computationally demanding optimization issues that are crucial in intricate industrial settings is being studied more and more. Recent studies demonstrate their competitive performance against well-established classical approaches for particular issue types and their feasibility in fields such as robotic inspection.
Optimizing robotic inspection paths for industrial components, particularly those with complex three-dimensional geometries obtained from Computer-Aided Design (CAD) models, is a noteworthy area where D-Wave’s technology is gaining traction. A three-dimensional version of the well-known Travelling Salesman Problem (TSP) is used to frame this task. The goal of the TSP, a famous combinatorial optimization problem, is to determine the shortest path that makes exactly one stop at a specified group of “cities” before returning to the starting point. Researchers apply realistic limitations like open-route requirements and imperfect network architectures by modifying this problem for robotic inspection.
You can also read Naoris Protocol: Post-Quantum Cybersecurity For Web3
Research has compared D-Wave’s quantum-based solvers, which use a hybrid strategy that combines classical optimization methods with quantum processing principles, to more conventional classical optimization tools like Google OR-Tools and GUROBI. Car doors, bumpers, toy bears, aircraft parts, and H-spheres are just a few of the five different industrial case studies in which D-Wave’s hybrid solvers (NL-Hybrid and CQM-Hybrid) have proven to be able to produce competitive solution quality while drastically cutting down on computation times.
Even if traditional solvers like Gurobi occasionally produce marginally better results, D-Wave’s solvers’ speed advantage makes for an attractive trade-off, especially in dynamic manufacturing environments where flexibility and quick response are critical. D-Wave solvers routinely produce solutions with AR values ranging from 0.62 to 0.94, indicating near-optimal inspection trajectories, according to the research, which measures solution quality using the Approximation Ratio (AR) metric and computational efficiency using the Run Time (RT) metric.
Fundamentally, quantum annealing is a metaheuristic technique that uses quantum fluctuations to determine the global minimum of an objective function. Quantum annealers work by first setting a system of qubits in an initial ground state and then letting it develop adiabatically to a final state that encodes the required solution, in contrast to gate-based quantum computers that use controlled qubits in non-binary states. Compared to traditional “hill climbing” techniques like Simulated Annealing (SA), this method makes use of the quantum mechanical phenomenon of tunnelling over potential barriers, which enables the system to escape local minima more effectively.
One of the most promising systems for using quantum characteristics for computation at the moment is D-Wave’s quantum annealers, including the Advantage QPU with the Pegasus architecture. The Pegasus topology, which is a major advancement over earlier generations, links each qubit to 15 others. Even though these annealers are not universal quantum computers and have a smaller range of applications than gate-based systems, they are especially well-suited for optimization problems, particularly those that are closely related to the Ising model or can be expressed as Quadratic Unconstrained Binary Optimization (QUBO) problems.
You can also read Fractional Quantum Anomalous Hall and Quantum Computing
The study also examines how well D-Wave’s hybrid solvers perform on a wider variety of optimization tasks in comparison to top-tier classical solvers like CPLEX, Gurobi, and IPOPT. By breaking down problems into smaller ones and executing some of them conventionally and others on the Quantum Processing Unit (QPU), these hybrid solvers overcome the scalability constraints of existing quantum computers. For example, D-Wave’s cloud-based Leap Hybrid Solvers automatically apply penalty terms for constraints, frequently combining quantum hardware with classical algorithms like Tabu search and simulated annealing.
D-Wave’s hybrid solver produced optimal results for Binary Linear Programming (BLP) with linear constraints that were on par with CPLEX and Gurobi. But the more variables it had, the longer it took to calculate, especially when it got close to its scalability limit. At D-Wave’s variable limit, the addition of additional constraints resulted in a significant rise in calculation time and a divergence in solution quality from conventional solvers.
D-Wave discovered suboptimal solutions, frequently two times off from the optimal, when solving integer/binary linear programming problems with quadratic constraints, indicating that it might become trapped in local minima. Nevertheless, for some variable counts, it sometimes solved problems more quickly than IPOPT and CPLEX.
These problems are particularly well-suited for quantum annealers because Binary Quadratic Programming (BQP), which theoretically mimics the Ising Hamiltonian, is where D-Wave’s hybrid solver shown its most promising performance. D-Wave showed a definite computational advantage in this particular scenario by consistently finding optimal solutions in less computational time than CPLEX (when capped at 1000s runtime) and IPOPT for BQP with 500 binary variables.
However, even with longer run times, D-Wave’s LeapHybridCQMSolver was unable to compete with classical solvers like Gurobi in terms of computational time and solution quality for Mixed-Integer Linear Programming (MILP), specifically applied to a real-world Unit Commitment problem in the power sector. This implies that although D-Wave may solve MILP problems, its computing advantage seems to be restricted to situations where the objective function or restrictions contain quadratic binary components.
Overall, although while Advantage quantum computers and D-Wave’s hybrid solvers are starting to compete with traditional optimization techniques, their strength is now only shown for a small number of problems, especially those involving binary quadratic structures. The sector is developing quickly, and D-Wave is unveiling upcoming topologies like Zephyr and constantly improving its gear. With further advancements, quantum computing has the potential to transform industrial optimization and yield enormous financial gains. Open access to research data encourages more cooperation and speeds up the creation of quantum-inspired solutions.
You can also read Spontaneous Symmetry Breaking Simulated at Zero Temperature
