Artificial Intelligence AI in Aerospace

Artificial intelligence (AI) in aerospace—is it here? Well, the answer is yes. AI is here and it is here to stay. So, how do we as an industry deal with the use of artificial intelligence, more commonly known as AI, in the aerospace sector? There are a couple of things to consider when discussing AI.

With no specific regulatory requirements covering or capturing AI-specific development in aviation, there are basic framework road maps that define the use. However, there are many standards that outline system requirements and development capturing these ideas and thoughts. Some systems currently use AI for analysis of big data, such as network security and aircraft health monitoring. So, let’s discuss analytics of big data.

Current aerospace systems produce large quantities of data for analysis, which cannot be completed in a timely manner by human intervention. With the use of a qualified AI tool, the analysis can be greatly reduced from a month or more for a single data set to mere hours. So, there are tools utilizing AI by means of algorithms to disseminate the data in a timely manner. The output reports of the analysis provide opportunities for safe decision-making. These tools can be set up to include old data with new data to establish new prediction model levels. These new prediction levels assist in planning maintenance actions before a system or component failure occurs. By utilizing these predictive AI tools, the aircraft is maintained efficiently for continued in-service safe operation.

A couple of concerns utilizing AI in the aerospace sector are factors such as safety, security, and trust. Let’s discuss these factors. First, safety: is AI safe? Regardless of the platform, safety is always of top concern. Safety of the ground crew, flight crew and the public. To meet these safety factors, are the aspects being implemented at the onset of system requirements? It is crucial for system requirements to include, in whatever manner, AI usage. If the requirements are not specified for AI, there are no controls at the beginning of a development to curtail unvetted usage. The system requirements set should outline the details of what type, the use case, and how the requirements set will be verified and validated. This is crucial to assure safety applications are instilled in the various activities.

“Current aerospace systems produce large quantities of data for analysis, which cannot be completed in a timely manner by human intervention. With the use of a qualified AI tool the analysis can be greatly reduced from a month or more for a single data set to mir hours”

Second, security: is AI secure? Continual technological advances require strict adherence to security measures. Technology security measures assurance utilizing AI, whether part of systems architecture or tool qualifications, are critical and must be defined within the initial requirements set. Lastly, Trust, Is AI trustworthy? This AI factor may be the largest hurdle of all to overcome. The developers, designers, and end users are using engineering best practices, standards and methods to assure AI can be trusted. However, public trust is the true test of whether AI will be accepted in aerospace.

It’s concerning, especially with the large amount of data to be processed and with the negative publicity AI has received recently, it is difficult for the public to have any trust that AI will not become intrusive. As with most things, the public can affect the outcome of products and/or businesses. If the public does not trust the product or the business, that item eventually will cease. AI does have a place in our lifetime and the aerospace industry.

The aerospace industry has always strived to provide a source of transportation in a safe and economical manner, whether transporting passengers or cargo. Technology changes over the past couple decades play into the ability of industry to provide safe and economical transportation. These changes have increased the amount of activity to maintain in this fashion. To meet this task and to maintain public trust requires systems and programs for predictive activity by fixing a problem before it happens. Predictive activities provide a method to improve reliability by planning maintenance activities ahead of time.