Artificial Intelligence (AI) is beginning to permeate every aspect of daily life, from garnering film recommendations on Netflix, through to fraud detection and cognitive robotics, and airport security is no exception.
Machine learning is key to supporting the development of self-teaching algorithms that allow airports to maximise the efficiency and safety of airport checkpoints.
Security screening processes generate large volumes of data, which are primarily used to identify and prevent security breaches. The exponential growth in computing power, combined with the spread of data analytics and the sheer amount of available data, now make it possible to develop AI algorithms, which deliver truly meaningful insights and improvements for both the screening and operations.
Automatic object recognition
Smart, adaptable algorithms are now being developed for the automatic detection of an expanding list of dangerous, prohibited and contraband goods and substances.
The application of deep learning algorithms for automated threat detection requires the availability of a considerable image database, categorised in threats and unsuspicious images. Deep learning algorithms scan this information to ‘learn’ which objects are potentially harmful and which are benign. These algorithms provide invaluable support for security operators, customs officers and other controlling authorities. This new form of detection reduces the burden on image analysts and increases efficiency and throughput. Delivering improved levels of both safety and security, the algorithms can identify potential threats and help combat the movement of unsafe, undeclared or illegal goods.
The concept of risk-based screening is designed to increase the operational efficiency of screening resources and processes as well as improve security levels.
Advanced data analysis can help airports ensure that their resources and measures are where they’re most needed, which ultimately makes for a smoother passenger experience.
Deep learning algorithms can combine a comprehensive set of passenger information to create a risk profile. This includes travel data, passport data, visa and biometric information, and flight behaviour patterns, as well as critical data from government and security services on criminal records, ‘person of interest’ files, and other relevant intelligence sources.
Artificial intelligence is able to efficiently pass through the data to identify key patterns and help inform the assessment.
With risk-based screening, integrated networked checkpoints could trigger personalised security measures in response to a change in threat levels.
Biometric identification is already being used at many airports around the world to create a secure and seamless passenger experience from check-in to boarding and is now making its way to security checkpoints as well.
This integration will be a key enabler to match passengers with their trays at divestment and allow for dynamic passenger differentiation during the screening process as part of a risk-based screening approach.
AI techniques have significantly improved facial recognition accuracy levels to reach an all-time high, which in turn has resulted in fewer manual interventions being necessary allowing processes to become much quicker.
We believe that these rapid improvements in AI technology along with the clear benefits associated with its use will continue to push its integration into checkpoint security and every other part of an airport’s operations.