Protiviti recently held a workshop with four large customers facing various challenges with distribution and logistics ranging from vehicle routing to warehousing. All attendees were interested in gaining an edge with quantum computing for such use cases in the near term. Over a half-day, Protiviti presented the basics of the “art of the quantum possible” in this sector, along with a live demo on a quantum computer of a vehicle routing proof-of-concept (PoC) we developed in-house. We rounded out the workshop with a design thinking session to get attendees brainstorming about what types of use cases they would like to tackle with our help.
This isn’t the first time we held such a workshop, yet the results continue to surprise and delight us. Sure, we spend time adapting academic papers that apply quantum algorithms to what we feel are practical use cases. But this lab approach does not fully capture real-world constraints quite like these interactive sessions. Hearing from our customers spurs innovation. Here are just some of the novel approaches we will be working on in coming months as a result of their feedback.
Vehicle routing
Our quantum hardware demo focused on vehicle routing, with constraints we added to make the problem more interesting. How are packages arranged in the truck? In what order are they delivered? These might be interesting constraints to showcase the power of an approach, but customers have other constraints that are mission-critical.
D-Wave recently showcased a vehicle routing problem where the team simulated delivering supplies to an area affected by disaster (see The Post-Quantum World episode here). The classic K-Means approach made the deliveries with 27 kilometers of driving, avoiding downed power lines, flooding and other artificial constraints. The quantum annealing approach, running on D-Wave’s hybrid system, routed the same deliveries with only 20 kilometers of driving. It’s easy to extrapolate the benefits of only doing 74 percent of the driving previously done in a day.
A word on quantum PoCs—they’re always best done on small subsets of production problems. There are still access bottlenecks to quantum systems on the cloud, and only so many spots in queues to go around. Using a subset of a problem makes it easier to tweak a quantum algorithm then show meaningful results quickly. We can then add constraints and scope until we reach the limits of a quantum device. It’s possible to analyze results and extrapolate how performance will improve as qubit counts and quantum volume grow in coming months.
Back to the customer constraints discussed in our workshop. One has a huge fleet of vehicles that would probably exceed the limits of quantum circuits we can run today. Luckily, we uncovered a natural subset to experiment with. The customer has a relatively small batch of electric vehicles in the fleet.
Imagine the possibilities. EVs have unique concerns. They have a certain charge level at the start of a day. We must consider the location of the available charging stations. What is the charge rate? How about battery degradation—do we cap charging to 80% or a similar value to encourage more duty cycles and overall battery longevity? Succeeding at a subset like this could lead to rolling out to fleetwide production systems, especially as fleets are expected to shift to EVs over time. No doubt some of these will be autonomous, which could add further constraints in the future.
Warehouse management and packaging
Discussing fleets of vehicles and their freight took our workshop attendees figuratively and literally to warehouse issues. Could we use quantum to identify upstream supplier risks in the supply chain? Experimentation is needed to map such a problem to qubits. However, it does seem likely that quantum could help forecast shipping lead times, costs and possible alternatives. As with the vehicle fleet, we’d want to box off the problem to a subset by choosing from the available data to get started, then expand.
For example, a popular proof of concept is the optimal packing of shipping containers or the bin-packing problem. Here, quantum algorithms have been used to make optimized decisions based on size, weight and value of the items. This basic approach can be easily modified to solve some of the other problems attendees brought up. One example is optimizing product stocking levels to increase space efficiency in warehouses. It’s not just Tetris-like stacking here. Other constraints include the expected shelf-life of goods and how often similar items are ordered by end customers.
Customers brought up concerns with 3D printers, and we noted that these can be thought of as miniature warehouses, too. They need consumables procured in an efficient way. And users must consider the order of print jobs, what a printer can handle and types of materials to store (based on strength or flexibility or cost). And how are finished items moved efficiently? It’s a miniature ecosystem where quantum can have an impact.
The warehouse location itself can even be optimized. This is an established problem with known tested classical computing solutions. Warehouses don’t move, so this might not be a dynamic enough problem to strain quantum computers, even if distribution centers may need to be relocated or reselected occasionally. The benchmarked solutions on the market make it a tempting PoC, though. It’s a rare opportunity to compare potential ROI from a quantum solution to tried and true classical. Taking advantage of existing benchmarks is, well, optimal.
Energy and ESG
We’ve only scratched the subatomic surface of how quantum will affect environmental, social and governance concerns. One use case of interest discussed was keeping energy available to end-users via distribution center power optimization and load forecasting. Attendees were also interested in using forecasted pricing and loads to take advantage of utility market prices in performing certain business operations for less.
The vehicle routing problem came up again in our ESG discussions. Obviously, having efficient routing of vehicles also affects the energy efficiency of shipping. Companies can save money by optimizing routes but could also improve how customers perceive the organization’s concern for the environment. Quantum may also adapt bin-packing algorithms to help improve how individual products are packaged in a box, minimizing material waste and weight.
While we haven’t reached benchmarked quantum advantage yet as an industry, use cases like the ones mentioned here give our team real-world problems to work on in the coming months. We’ve already developed a working PoC around the EV use case and will be experimenting with the others next. Our hope is to keep stimulating the innovation process with brainstorming sessions with customers. What problems are you hoping to solve with quantum computing?
Listen and subscribe to The Post-Quantum World podcast wherever you listen to your favorite podcasts. Visit our quantum computing website to learn how we are helping clients get post-quantum ready or contact us for more information.
