A common problem at all airports is passenger safety in an environment of automated products for baggage handling. The main issue is the lack of understanding by the public of the inherent risks involved when interfacing with automated equipment. Even though all equipment is mitigated with machine guards, safety devices, Operator oversight, and warnings, there is still a huge risk of injury by misuse. The misuse is not intentional, but the implications can expose the individual to a high risk of injury. This applies to all, young or old, with a particular risk with unsupervised children.

A strange phenomenon occurs at Airports, where people feel that they are in a relatively safe environment and relax their normal sense of security and unintentionally compromise their personal safety. Sitting on moving equipment, climbing over barriers to reclaim their luggage, children playing on baggage carousels, are all common occurrences in the daily operation of an airport.

So, what can we do to mitigate the risks even further? Given that legislation already applies the minimum standards of safety, what extra layers of protection can be provided to guarantee passenger safety, even though they are oblivious of the risks that they are taking.

With the emergence of the second dawn of Artificial Intelligence (AI), a new way wave of options is available with the capabilities of machine learning, this gives an opportunity to use AI in a layered approach to safety and security.

This is where a little creative thinking and partnering with like-minded Engineers, allowed us to join to together to solve this particular problem. Working with a new start-up company called Machine Learning Technologies, who were creating different AI models, we worked collaboratively to design and trial a Human Detection system using a combination of AI and basic imaging equipment (camera’s). This required several AI models to be created: One for discriminating between the world and a Human form, another for Human Gesture identification, and lastly one to determine baggage from everything else.

Of course, using AI requires a huge amount of training the different models to reduce the number of false positives (mistakes) by the models and to increase the reliability of detection and identification of individuals, items, and bags. This was completed over a period by collecting and processing thousands of images of People and bags. This initial training was bolstered by a real-life trial in London Stansted’s busy Arrivals Hall. This accelerated the model to a point where the false positive alarm rate hit zero, making the AI competent in the discerning people and bags.

Though, AI cannot be solely used for safety purposes, using a layered application of deterrence, passive systems (guarding), and active systems (Safety devices and AI) enables AI to be efficient in detection and verified by the other devices.

John Golding, Owner of MLT, on the future of this technology explained the possibilities of the technology,

 ‘Building on the success of our recent trial, we have now incorporated other cutting-edge developments into our platform. The introduction of real-time luggage IATA classification and passenger heatmaps within the arrival's hall will further our mission to create a seamless, secure, and highly efficient airport environment. These features represent our commitment to continuous innovation in response to the evolving needs of modern forward-thinking airports.’

Where does it stop? With the abundance of data collected daily by companies and the flexibility of AI, machine learning, and deep learning machines, the applications are endless when it comes to the ability of AI supplementing our businesses and leisure time. So the second dawn of AI is bringing a new way of looking at complex engineering problems and using innovation to simplify the solutions and produce better results of a repeatable nature.

By Steve Radford EngTech MIET,  Head of Engineering – London Stansted Airport