In a forest of English acronyms, the Marche Polytechnic University has announced the testing of a system that, using radar, microphones, cameras and electronic intelligence, promises to signal the imminent passage of emergency vehicles to drivers.
The goal of the project called “Smart multimedia MONitoring systems for advanced HMI technologies in living environments” (hence MOHMI) is to incorporate hardware and software in the cars of the future which, working in unison in different driving contexts, will be able to support the driver in perceiving with even greater clarity the events outside the vehicle. The studies are part of the wider MIRACLE project – another Anglo-Saxon title which translated into Italian stands for Innovation and research of the Marche region towards connected and sustainable habitable ecosystems – and at the center of the countless analyzes, there is the role of artificial intelligence in capturing and include external sound events, such as the passing of an ambulance with sirens blaring. To the electronic ears, the system then adds an electronic eye (artificial vision system) used to pick up the reactions of the driver in the immediate onset of the sound of a siren. If the driver has not heard or perceived the input, the system informs the occupants of the passenger compartment and in particular those behind the wheel to pay attention.
The set of radars, cameras and external sensors to “read” the events on the road, are technologies used by car manufacturers since the beginning of the new millennium (in some cases even earlier) and in the constant increase of active safety and passive, electronic eyes and ears have evolved in several directions in order to acquire data inherent to real driving scenarios, with ever greater precision. The microphones used by the studio are in fact designed to withstand the elements, and positioned to primarily capture the sounds coming from the back of the car, which are generally more difficult to perceive.
Collecting the many sounds of traffic, the algorithms of the project recognize and separate the sound inputs, indicating at the same time to the driver the type and direction from which they come. The main concept is to help the driver immediately understand which direction to look even in the event of a lot of noise and with an accuracy of at least 99 percent. The researchers contextually analyzed the
behavior of drivers upon arrival of the rescue vehicle, using the semi-anechoic room of the A3lab group (Department of Information Engineering) at the Polytechnic University of Marche. The semi-anechoic chamber simulates a closed environment where sound is absorbed by the walls and the one supplied by the university housed a car for the various sound tests explored by the studio. The processing of the reactions within a scientific and controlled environment, made it possible to evaluate with empirical data how the degree of attention of the driver suffers many distractions while the sensors of the Dynamic Vision System (DVS), were able to analyze in real time and successfully through appropriate neural networks, the opening and closing of the eyes, and the movements of the head and mouth. Finally, two further video cameras calculated the depth allowing the engaged devices to interact and learn the environment better by checking, for example, if the driver in question observes the rear-view mirrors when the sirens sound. The cockpits of the future are therefore preparing to be more and more X-ray spaces where in the name of safety and the reduction of accidents, the deeds of the human being will be analyzed 24 hours a day.