The AI solution that sorts through billions of possible last-mile routes for the best one

There is a reason artificial intelligence (AI) proves itself to be an effective tool for route optimization.

"There are 620,448,401,733,239,000,000,000 possible routes between 24 locations," said Paul Hart, Head of Logistics at AI technology company Satalia. "Searching through all of those possibilities, it would take a computer 2 trillion years to find the optimal route. And that doesn't include the complexity of shift patterns, vehicle capacities or any other real-world constraints."

Started by academics from University College London (UCL) in the UK, Satalia has used AI to solve difficult problems for 12 years. The company, which uses HERE maps and data for its calculations, helps big names including major supermarkets across the world.

“30 years ago, if you ordered something from a catalog, you were looking at one month for it to be delivered," Hart said. “Now, it is usually the next day or even the same day. Customer demand is the push for these systems."

Algorithms detect the best available route in seconds as customers pick their delivery slot from a website. The difference between grocery deliveries and many other goods is that the customer gets a one-hour time window from the moment of ordering. Customers ordering online for non-food products usually only get an estimated time of arrival (ETA) on the day itself.

“That is why we needed more advanced systems, using the cloud and more advanced optimization algorithms," Hart said.

From HERE to there

A detailed set of information is required for the algorithm to make an accurate calculation. It must have the addresses of everyone an individual truck will deliver to during the timeframe, the distance between each and the time it will take, along with the type of vehicle used.

Average travel time is used to calculate the route since it will be decided ahead of the day and time of delivery.

“HERE provides the map data that our routing engine then runs on to create this matrix — which happens in real-time for customers," said Hart.

There is a surge in demand for delivery slots close to Christmas, as customers rush to the sites of stores to get their groceries delivered on time.

“That then puts a big demand on our service to be able to make sure that we can meet those demands," he added. The company uses the cloud to be able to scale its service up during these peak times.

The results speak for themselves. One company estimates it has saved 11.2 million miles since using Satalia and another reports a 9% increase in customer happiness. Satalia has helped one of its customers reduce overtime by 19%, meaning drivers are able to take breaks when they need them, even during the busy holiday season. There are also sustainability benefits.

“By having more efficient routes, you not only save time but also save emissions," Hart said.

Some of Satalia's most interesting offerings are a result of converging different areas of their AI business together. A workforce offering that matches people to tasks can provide the right drivers at the right time to run logistics operations smoothly, for example.

Micromarketing plans to push specific products can be based on what companies have in storage, helping to shift goods and clear warehouses through this targeted promotion.

“We're very much industry agnostic," Hart said. “We specialize in those mathematically difficult problems which require AI to reach the most efficient solution."