We frequently examine how artificial intelligence (AI) is used in specific industries and sectors on our blog. Today, we want to discuss artificial intelligence in aviation. How this intelligent technology helps in building autonomous drones and aids diverse aviation and travel companies in offering better products and services? Let’s find out!
AI in aviation is a broad topic, as this technology can be used with ease to support air transportation, travel, and air force companies both in the skies and on the ground. For starters, let’s take a look at how AI in aviation is changing this vast industry. First off, we are going to talk about ATM (Air Traffic Management) and ATC (Air Traffic Control).
Next, we will see how AI is supporting aviation companies on the ground, especially regarding their services’ management. Lastly, we are going to examine autonomous aircraft, primarily drones.
Moreover, we have a couple of interesting examples of the sector leaders. How are companies like Airbus and Lockheed Martin using AI in aviation? If you’re an aeronautical enthusiast or simply work in this fascinating sector, we strongly recommend you read on. We have some fantastic news and use cases that will change your perception of modern aircraft. Let’s take off!
AI in aviation: ATM, FMS, and ATC
Until recently, air traffic management (ATM) and control (ATC) were almost exclusively based on human work and experience. After all, since the situation up there is constantly changing (just to mention different weather conditions and traffic), a human touch is indispensable. But is it really?
More and more often, companies and airports all over the world begin to realize that AI in aviation comes with some significant benefits. As it happens, tasks like flight planning, flow management, and safety assessments can be, at least to some extent, automated.
The reason is simple–the vast majority of scenarios and situations in the sky can be predicted and handled automatically. Four words–big data and machine learning:
- With big data, aviation companies can train their ML algorithms to take various variables and data sources into consideration. This way, the intelligent ATM applications can take weather conditions and air traffic into consideration and make appropriate decisions based on these two critical data sources.
- With machine learning, you can make your algorithms more and more effective over time. After initial training, they get better and better at operating in real-life conditions.
We could say that automated ATM and ATC result in improved predictability and efficiency. For example, these two technologies can be used to calculate the optimal routes that will allow airlines and air transportation companies to save time and fuel.
Eurocontrol, a pan-European, civil-military organization dedicated to supporting European aviation, predicts that the usage of intelligent big data analysis and machine learning can result in improved efficiency, and the potential gain can reach up to 30%! That’s something worth fighting for!
FLIGHT MANAGEMENT SYSTEMS (FMS)
We can expect that, in the near future, every FMS will be based on these intelligent technologies. Shortly put, an FMS is a specialized computer system that automates a wide variety of in-flight tasks, reducing the workload on the flight so that there’s no need to take engineers or navigators on board. You can think of it as top-tier GPS teamed with aircraft monitoring capabilities. FMS takes care of the flight plan, navigation, and plane’s position. It’s indispensable in modern aircraft. Therefore, even light planes are equipped with it.
The FMS of the future will work based on various artificial intelligence solutions. The idea is simple–to make the most of available onboard and external data, primarily concerning weather conditions, route efficiency, and ongoing air traffic. To show you what’s about to happen shortly, let’s use a simple GPS example.
When you’re driving a car with Google Maps guiding you towards your destination, Google’s algorithm is showing you the best way. What happens when there’s a sudden traffic jam? The map instantly shows you an alternative route that will allow you to bypass it.
A similar solution can be used in modern FMS. The real-time data concerning weather storms, turbulences, increased air traffic, or other adverse circumstances can be used to recalculate the route and direct the plane into a different, optimal direction.
AIR TRAFFIC CONTROL
Now, let’s talk about ATC for a few moments. When it comes to air traffic control, the main objective is to keep everyone safe. ATC is typically managed from the control tower, where dozens of ATC specialists guide and communicate with the nearby planes and manage their landing and taking off.
The amount of data flowing through each ATC entity is simply unimaginable. That data can be easily used to train advanced machine learning algorithms that will take all the variables into consideration. Naturally, this doesn’t mean that the ATC work can be fully automated. At least for now, human experts are still necessary in the control towers. After all, there are some critical elements that need to be taken into consideration:
- Landing sequence
- Individual pilot behavior
- Accidents and failures
- Weather conditions
- Sudden situations (like emergency landings), etc.
But, with ML on board, the job of ATC specialists can be massively simplified. One of the companies working on such AI-fueled ATC systems is Swedish LFV. They collaborate with IBM to create an ATC system called Advanced Autoplanner. Currently, they have the first proof-of-concept, which is a model that provides air traffic control instructions in a Swedish en route sector .
If you’d like to see how LFV’s prototype works, here’s a video for you:
When it comes to AI in aviation, ATM and ATC are the very first aspects of modern aviation that we wanted to talk about. However, artificial intelligence in aviation is a much broader subject! Some of the applications are also commonly used in other sectors (like, for example, dynamic pricing). But let’s see how these applications can be used in the aviation and travel sector.
Other applications of AI in aviation
If you read our blog, you already know that generally speaking, AI has two major purposes:
- To reduce costs
- To improve efficiency
The airline sector is no different. Here, we’re fighting for exactly the same cause. So, what can be done to improve the operation of airlines and other aviation companies? Let’s take a look at some examples:
AI in aviation: DYNAMIC TICKET PRICING
If you’ve ever booked a flight ticket, you know it’s an experience like no other 🙂 The same flight can have different prices depending on your flight comparison engine. Prices also differ depending on the departure time, destination, flight distance, and the number of available seats. The cost of the same ticket can change by the minute.
How is that possible? Well, that’s because airlines use something called dynamic pricing. It’s a technique of adjusting prices based on the current situation to the most profitable levels (of course, from the airline’s standpoint, not yours).
Frequently dynamic pricing algorithms use intelligent solutions like machine learning and big data analysis. And although you may not be a huge fan of this solution, the fact is, it’s the most common AI application in the aviation world.
AI in aviation: DELAY PREDICTIONS
Yes, we have to mention weather conditions for yet another time today. Delays are unfortunately common, and they depend on dozens of different factors. Modern ML-based applications can help airports and airlines all over the world predict delays and inform passengers as quickly as possible. This way, the aviation companies can also significantly improve UX (User Experience), as customers will have more time to re-book their flights or make other arrangements if necessary.
AI in aviation: FLIGHT OPTIMIZATION
Partially, we’ve already discussed this application. Modern ATM systems enable airlines and air transportation companies to set optimal flight routes. This way, they can lower costs, save time and fuel.
AI in aviation: CREW SCHEDULING
In essence, it’s nothing more than a typical workforce management (WFM) feature. You can read more about WFM in this blog post. However, when it comes to crew scheduling, there are several elements that have to be taken into consideration:
- Legal and contractual requirements
- Given employee’s qualifications and certifications
- Personal preferences
Airlines have to deal with complex networks of employees, and that’s including flight attendants, pilots, engineers, and other specialists, making necessary pre-flight preparations.
Intelligent WFM systems help airlines in scheduling crew members for every flight without unnecessary complications or delays. This way, each flight has an ensured number of crew members and can go as scheduled, and potential errors are reduced to a minimum. It’s also the best way to use the full potential of every crew member.
AI in aviation: PREDICTIVE MAINTENANCE
Predictive algorithms are, at least in our opinion, one of the most impressive aspects of the entire AI industry. With these intelligent solutions, airlines can predict flight delays, potential complications, but also necessary repairs and maintenance procedures.
Like all other machines and vehicles, aircraft need appropriate maintenance so they can remain fully functional and safe. In one of our past blog posts, we talked about something called a digital twin. A digital twin is an exact digital replica of a specific machine or device (in our situation, aircraft).
Now, thanks to IoT (Internet of Things), companies working with digital twins can replicate the exact state of the physical object in their applications. In other words, companies managing planes and conducting repairs can have the necessary insight into each machine and assess when specific repairs and maintenance are necessarily based on the plane’s (and its crucial components) condition. It’s a terrific way to optimize work, save money, and make sure each plane is always in excellent shape.
Furthermore, thanks to IoT, machine learning, and predictive algorithms, companies managing aircraft can predict potential failures and glitches on a plane before they actually happen, which also contributes significantly to the time and money savings. We don’t have to say that incorrect repair or delay in the plane’s maintenance can result in hundreds of thousands of dollars of additional expenses.
These are the most popular elements of AI in aviation. But as we already told you, aviation is so much more! For instance, we have drones and UAVs (Unmanned Aerial Vehicles).
AI in modern drones
Crewless vehicles are more and more popular. Also, they are more intelligent and autonomous than just several years ago. And the word “autonomous” is key here, as AI is what gives drones and other UAVs full autonomy.
Obviously, some of the UAVs have to be controlled by a remote pilot, located typically tens or even hundreds of miles from the vehicle. But there are also fully autonomous aircraft that operate on massive amounts of data and work entirely on their own thanks to machine learning and other AI-related technologies.
When we’re discussing drones, it is vital to mention some critical aspects of their activity, and these are state estimation, control, mapping, and planning. How can AI in aviation help with these four elements?
STATE ESTIMATION AND CONTROL
In general, that’s the ability to estimate the current drone’s position. Typically, drones can localize themselves by using sensors to measure environmental features and then by registering the measurements against a pre-existing map. Sometimes they are equipped with GPS or SLAM systems (Simultaneous Localization And Mapping).
AI comes in handy, especially in challenging or even dangerous circumstances. With ML and other AI-fueled technologies, drones can quickly adapt to the ever-changing situation and issues like, for example, a broken propeller. Sometimes even deep learning is used to help UAVs cope with difficult situations without human assistance or supervision.
As we can read in the “Estimation, Planning and Mapping for Autonomous Flight Using an RGB-D Camera in GPS-denied Environments” paper: “map updates and global pose updates are not required at a high frequency and can therefore be processed on an off-board computer”.
However, currently, drones are equipped with top-of-the-line mapping systems that help them understand the environment they’re operating in. These systems comprise various sensors, radars, and lidars. Mapping is typically supported by advanced computer vision (CV) systems that work similarly to the human eye. The camera registers the picture in real-time, and advanced analytics systems process incoming data.
Lastly, we ought to mention path planning. Our hypothetical drone already knows where it’s located (state estimation), how to fly in a specific direction (control), understands the environment it operates in (mapping). The last element is the way to reach the destination. And this is what route/path planning is all about. AI-fueled planning allows drones to devise the most efficient and quickest way. This is where another advanced technology called RL (reinforcement learning) steps into play.
With RL, the algorithm can learn and make necessary calculations in real-time, without the need for additional training. RL is also frequently used in collision avoidance, primarily in the environment where many drones operate in a limited space. This way, drones can operate on the battlefield or in the forest without the risk of hitting a tree. This technology also allows them to reach the destination successfully.
Examples of companies working with artificial intelligence in aviation
Perhaps, at this point, you’re thinking about the real-life applications of AI in aviation. In fact, almost everything we talked about in this article is already possible or will be in the near future. We already have the necessary technology and know-how. Therefore, we can expect that, in the coming years, AI in aviation will flourish.
To show you how far we are in the development of the necessary solutions, we want to use six exciting examples of companies working with or on artificial intelligence in aviation. How they use this technology to transform the industry? And what are their main objectives? Let’s find out!
It’s a US-based AI company specializing in vision AI, especially vision inspections… They are working on a system that allows drones to successfully navigate through trees, buildings, and other obstacles in crowded environments. According to Neurala’s CEO: “AI can make decisions faster than a human in situations like collision avoidance. If a human takes a few seconds to react, at 70 mph it is already too late”.
AI-fueled drones don’t have such limitations and can act almost instantly. With this idea in mind, they’ve created a system called Neurala Brain, also referred to as onboard AI. This system can be easily implemented even in small-size drones and UAVs.
It’s another AI company from the United States. Although they concentrate on various areas within AI, self-driving models are one of their main areas of interest. Their solutions can be applied to UAVs and cars and help them in:
- Video object tracking
- Semantic segmentation
- Detecting traffic lights and street signals
Thanks to their solutions, ML-powered drones can identify and map all the relevant objects like houses, cars, traffic lights, and many others and use this knowledge in their path calculations.
APPLIED AERONAUTICS: ALBATROS
As we can read on the Applied Aeronautics website: The Albatross can fly for up to 4 hours, boasts a 10kg MTOW, and is entirely open and customizable, making it ideal for boundary-pushing research development projects as it is for commercial operations. It can travel for over 100 miles reaching top speeds of 90 mph, carry up to 4.4 kg of additional payload and fly for up to 4 hrs on battery power.
Such a light aircraft can be used in many challenging tasks where aerial surveying is necessary. And because this UAV is customizable and affordable, companies worldwide can use them for their purposes.
Boeing is a perfect example of a company that uses AI on the ground to improve their production processes. In early 2019, Boeing informed the market that they are using AI in aviation to drive even more efficiency from precision automation equipment assembling aircraft in South Carolina, US. They are focused on increasing productivity, primarily when it comes to fuselage section assemblies for their 787 Dreamliners. Their ML-based machines perform drilling/countersinking, sealant application, fastener insertion, collar swaging, positioning sensors, and early fastener feed.
Boeing’s major competitor also eagerly uses artificial intelligence in aviation. They are, however, focused primarily on autonomous air travel. Airbus sees the major advantage of AI in aviation in fuel savings, lower operating costs, and support for pilots during their work. Currently, they are working on four crucial applications of AI in aviation:
- ATTOL: This abbreviation stands for The Autonomous Taxi, Take-off and Landing (ATTOL). The goal of this project is to use computer vision to enable commercial aircraft to navigate and detect obstacles during taxi, take-off, approach, and landing.
- Fello’fly: This project is focused on the usage of wake-energy retrieval for commercial aircraft. Airbus closely examines how planes flying together (inspired by migrating geese) could boost the environmental performance of aircraft and help decrease fuel consumption. Currently, Airbus estimates that this technique has the potential to generate 5-10% savings per trip.
- Disruptive Cockpit (DISCO): It’s an enhanced cockpit that is designed to enable single-pilot operations for new aircraft.
- Wayfinder: Here, Airbus concentrated on building scalable autonomy systems that power self-piloted aircraft applications, from small urban aerial vehicles to large commercial planes.
Due to their military pedigree, this company is concentrated on using AI in aviation to support intelligence, surveillance, and reconnaissance (ISR) operations, especially when standard communication between systems is no longer possible because of adversary activity. Their ISR solutions harness the power of AI to enable safe access to contested environments and gather the critical intelligence necessary to make strategic decisions. Their experimental ISR system has been tested in mid-2020 at the US Air Force Test Pilot School at Edwards Air Force Base.
To sum up, AI in aviation is a fascinating and continuously developed field of expertise. We can expect to see many more amazing applications of this technology, both in commercial airlines, air forces, and transportation vehicles. Thanks to artificial intelligence, airlines, and carriers can optimize their routes, save time, and improve UX. If you are interested in how AI can help change your industry–drop us a line! At Addepto, we are experienced in working with companies representing many different sectors. We will gladly help you as well!
 Community Alternatives to Luton’s Flight Path. Technical and Safety. URL: https://calflightpath.org/technical-and-safety/. Accessed Apr 8, 2021.
Eurocontrol. About us. URL: https://www.eurocontrol.int/about-us. Accessed Apr 8, 2021.
 LFV. AI enhanced Air Traffic Control. URL: https://www.lfv.se/en/about-us/innovation/ai-enhanced-air-traffic-control. Accessed Apr 8, 2021.