AI-Powered Route Optimization for Air Cargo

Back in 2018, a trio of start-uppers came to Alaska Airlines at the invitation of their friend. Their job was autonomous driving. They came to explore if the artificial intelligence models they develop can be used in aviation route optimization. What they saw there was surprising…

In this article, we shall discuss how the ending of this story helped an aviation company save millions and how artificial intelligence can improve route optimization in air cargo.

Let’s get started!

How AI helped to reinvent the dispatching processes at Alaskan Airlines

The three start-uppers were the founders of Airspace Intelligence, Phillip Buckendorf, Kris Dorosz, and Lucas Kukielka. Here’s how they describe their first impressions at Alaska Airlines: “We expected to see science-fiction-like systems, like we know from movies.” In fact, they saw dispatchers sitting at old-fashioned IBM green screens, studying weather charts printed out on paper using software that was 20 years old.

Alaska Airlines’ operations control center, called NOC, or the Network Operations Center, contained about 100 dispatchers continuously monitoring weather forecasts, news outlets, and FAA websites to gather information needed to build routes properly.

Here’s how Peter Saleh, the head of corporate development, who invited the AI software developers, describes the process: “You have a tab open for the Weather Channel, a tab for CNN… So it’s just click, click, click. If you look at a dispatcher, they have, like, 19 tabs open that they’re flipping.”

In the next two years, the group studied carefully (basically camping at NOC with sleeping bags) how the dispatching processes operate and how AI can improve the process. As a result, the Flyways software was created.

It changed a typical dispatcher’s working day greatly. Now, the Flyways software funnels the information on the web and presents it in a user-friendly interface. As the routines are automated, the dispatchers have more time to plan. Also, as the dispatcher is busy planning a route (typically, a dispatcher has 20 routes assigned), alerts on the possible improvements of other routes pop up.

How does AI for route optimization work?

Artificial intelligence works with a bigger amount of information than a human can do. AI algorithms rumble through historical or real-time data and bring out the findings. Although the principle of work is single: the super-fast analysis of huge amounts of data, the outcomes can be used in different instances. 

How is AI used for route optimization?

In fact, artificial intelligence has multiple applications in air cargo route optimization. First of all, machine learning and deep learning are used to analyze historical data and see patterns in traffic load. Secondly, AI can process real-time data to make exact predictions. Also, real-time data analysis helps respond quickly to incidents. Let’s review each of the instances closer:

Machine learning

Aviation companies have collected tons of high-quality statistical data for decades. Yet, most of the documentation was previously stored on paper and was inapproachable for digital analytic tools. Now, with the rise of AI-powered optical character recognition, things have changed. The newer OCR tools trained with deep learning algorithms can read and process documentation in handwritten, printed, or typed format, as well as images. After, this information gets sorted, grouped, and systematized as needed. New OCR technology unlocks years of historical data for analysis on behalf of AI tools. Interested in the possibilities of AI-powered OCR technology for your business? Check the article “OCR in Document Management for Aviation Companies.” Deep learning algorithms process historical data to spot previously unknown patterns. An algorithm can check the history of winds in a certain area, as well as flight duration under different weather conditions, etc. By spotting historical trends, AI offers the most efficient scenarios and predictions. For example, if a flight is scheduled to start with a favorable wind, the flight duration can be reduced by up to seven minutes.

Predictive analytics

Air cargo route planning depends on a huge amount of data. This includes whether there is any military training in the area, if there will be other planes landing there if some of the planned flights were canceled, etc. The analysis of all the data is quite complicated. Sometimes, two planes are scheduled to come to one place at the same time. Then, one of the aircraft gets ordered to circle around a certain area, which uses additional fuel consumption. So, how do AI tools help to make predictions for better routes? AI system scraps all the data in one place. Then, it analyzes all the data, which humans can sometimes skip. Then, based on a huge amount of real-time data, AI makes accurate predictions to avoid flight overlays and other similar situations.

Real-time data analytics for incident-caused data adjustments

AI can process data in real time. Therefore, AI tools can send alerts on incidents right as they occur. This gives a dispatch officer time to make adjustments faster and more accurately. For example, if an airport gets closed due to an incident, the flights scheduled to land there may be redirected faster.

The benefits of artificial intelligence in air cargo route optimization

Cost savings – by making operations more efficient, air cargo companies cut unnecessary spending. For example, Swiss Airlines managed to save $5.4 million USD in a year by optimizing about 50% of their flights. Lufthansa also uses AI analysis to switch planes effectively as customer demand changes. Thus, they use smaller planes for flights with lower demands. That saves costs on fuel and maintenance. Also, the AI tools determine whether the planes have leased engines or those owned by Lufthansa. Thus, they prefer planes with company-owned engines, which also save costs.

Environmental impact – by making flights more efficient and shorter, the airlines save on fuel and maintenance costs and reduce carbon emissions. For example, on average, by saving 2.7 minutes per flight, Alaskan Airlines avoided 6,866 metric tons of carbon dioxide emissions. Also, since Swiss International Air Lines has adopted artificial intelligence and Google Cloud technology to enhance its flight operations, it considerably cut carbon dioxide emissions. For example, their fuel consumption per person per 100 km decreased by 28% compared to 2003 and was 3.15 liters in 2022. It was less compared to 4.1 liters in 2021. In 2022, the Swiss managed to save 8,700 tonnes of CO2. By partnering with Google Cloud, Lufthansa Group plans to reduce up to 50,000 tons of CO2 emissions yearly.

Customer satisfaction- With flights coming on time and incidents managed promptly, air cargo clients trust their delivery vendors more. This doesn’t just lead to repeated contracts but also improves the brand’s reputation and serves as an additional marketing channel since satisfied clients are more likely to recommend the vendors they are happy to work with.

Better decision-making – correct decisions are based on the data available, and since AI for air cargo route planning increased the amount of data analyzed, it became easier to make the right decisions. According to Christian Most, senior director of digital operations optimization at Lufthansa Group, their human operations controllers use AI-proposed scenarios in 90% of their decision-making tasks.

Summing things up

So, it seems that the use of AI in air-cargo route optimization has passed over the starting phase and is confidently heading towards continuous development and deeper adoption. Based on the fact that the current partnerships between aviation and tech are young (Alaskan Airlines has been working with Airspace Intelligence’ since January 2021 only, and Lufthansa started working with Google and IBM in 2020) and the first results are really impressive, we can make the confident prediction about more aviation companies joining AI-based experiments.

Do you have an idea to use AI in your aviation business but lack the technical expertise to give it a proper evaluation?

Contact eNest for a free talk with our experienced AI specialist! We will answer your questions regarding the technical specifications of your projects, like how many AI software developers it will need and which technology to use, or we will provide you with an approximate cost and timeline of your potential project. Book a call now!

Meet a Expert

Have a question?

Speak to Data Scientist
Jagdeep Chawla

MS in Data Science
NorthWestern Univeristy, Illinois

Share on

Let's Talk Strategy