Developed in Pisa, AISAFETY monitors operators and machinery and intervenes in real time on production processes before critical situations arise.

La job security It is now at the center of a structural change. Transforming it from a reactive to a predictive approach is the same catalyst for efficiency—creativity—that is simultaneously revolutionizing all major professional and economic sectors, as well as, of course, artificial intelligence.

The fact that automation is no longer a distant theory but is already integrated into recruitment, planning, surveillance and security systems is also demonstrated by the growing space that the subject is taking up in the institutional comparisonToday, the uses of artificial intelligence and other technologies applied to worker safety are truly numerous and are finding increasingly widespread implementation, including wearable devices, sensors, AI-enabled cameras, exoskeletons, automation-assisted technologies, and predictive analytics systems.

One of the most recent and interesting developments in the predictive field comes from Italy. More precisely, from the University of Pisa, which presented AISAFETY, a supervision system designed to prevent machine-related accidents before they occur. This module can also be integrated into existing systems, reducing the risks associated with errors and tampering so that, as mentioned, Roberto Gabbrielli, professor of Mechanical Industrial Systems at Unipi and coordinator of the project, tragedies such as that of Luana D'Orazio, the 22-year-old worker and mother of a child who died on May 3, 2021 in a textile factory in Montemurlo, in the province of Prato, must never happen again.

Workplace accidents and persistent critical issues: why prevention alone isn't enough

According to INAIL data, in 2025 the number of victims at work in Italy was 1.093; numbers substantially in line with those of the previous year. This is a sign that the risk is not decreasing but rather being redistributed: while deaths occurring during work have decreased slightly, those related to commuting have increased.

Even the same people think about not changing sectors involved, with construction, manufacturing, transport and commerce remaining the sectors where the highest death toll continues, confirming critical issues that persist over time and that prevention alone struggles to contain.

Another delicate element concerns the profile of the victims, which is concentrated above all in the more advanced age groups of the active population, in particular between 55 and 64 and between 45 and 54, confirming in this case how experience is not in itself a sufficient guarantee to reduce risk, especially in the most exposed operational contexts.

Occupational safety, in addition to fundamental prevention measures aimed at eliminating risk and protective measures designed to reduce its impact, must be consciously integrated with technology. The adoption of advanced digital tools, in fact, does not replace the existing regulatory and organizational system, but rather strengthens it by introducing an additional capability: read the risk as it is forming.

The new supervision system combines sensors, cameras and AI for worker safety.

It is precisely from workplace deaths – due to carelessness or production dynamics in which safety is not the main focus – that AISAFETY was born. And although the final goal is still far away, the system of sensors and artificial intelligence developed byUniversity of Pisa demonstrates that the technology is already available and actually usable in real production contexts.

Errors or tampering can in fact turn into potentially fatal accidents in just a few moments: a minimal margin where AISAFETY acts by monitoring potentially dangerous situations in real time and intervening directly on the functioning of the machinery, to arrest them before the worst happens.

The entire system is based on this logic, combining sensors, cameras, and artificial intelligence to continuously monitor the interaction between operators and machines. A demonstration plant, consisting of a lathe and a robot, was also built at the Navacchio Technology Center, where AISAFETY was tested under operational conditions. The platform, as mentioned, is also designed to integrate with existing systems, without requiring a complete reconfiguration of the production systems.

It also explains it Paul Nepa, professor of Engineering in the field of Electromagnetic Fields at the University of Pisa:

"In the industrial sector, many injuries to machine operators are somehow related to user behavior. AISAFETY can also be integrated into existing machinery to reduce the risks associated with operator errors and tampering and allow them to work safely."

Three technological levels to read in real time what is really happening around the machines

To build the system, the researchers integrated multiple technological levels that work together in the same environment. A radio frequency system allows for real-time location of people and objects in the workspace, while artificial vision observes and analyzes what's happening around the machine and its operation. An artificial intelligence module then processes this data and identifies any risk situations related to errors or potential tampering.

"In the event of a risk, the Artificial Intelligence module activates shutdown commands on the production system to place it in a safe state. The supervision system also sends notifications to workers on their smartphones, advising them not to proceed with production to reduce the risk to which they may be exposed."

adds professor Roberto Gabbrielli.

The AISAFETY project (“Integrated smart system based on artificial intelligence for the management of operator safety in production processes”), developed together with the University of Perugia and the CNR, was co-financed by INAIL within the BRiC calls.

The other side of the coin, between a regulatory vacuum and rules to be built through training

While on the one hand innovative solutions such as AISAFETY clearly show that technology is already able to intervene on the risk in real time, on the other hand the overall picture remains unclear. far from full integration.

The analysis of the Ministry of Labor's report, along with the National Labor Inspectorate's in-depth analysis, clearly speaks of a double-edged sword in this regard. The same technologies that prevent accidents—from predictive maintenance systems to sensors that monitor fatigue, proximity, and environmental conditions—also introduce new risksAmong these, for example, the opaqueness of algorithms, which makes machine decisions less transparent, and the risk of excessive reliance on automated systems, which can reduce the capacity for human intervention in critical moments.

To this, as happens in the initial phases of all new technologies, there is added a regulatory node The issue is far from resolved. Risk assessment, required by Legislative Decree 81/2008 as a dynamic process and the direct responsibility of the employer, cannot be delegated to automated systems, while the use of monitoring technologies raises sensitive issues related to privacy and remote monitoring of workers.

A balance, at least the current one, where artificial intelligence does not replace prevention and protection, but redefines them, moving security towards a more proactive and predictive dimension which, however, requires rules and responsibilities still to be built.

Here are three insights that might interest you:

Fall arrest systems: with the new standard, safety becomes measurable.
Fraido: The New Frontier of Safe Clinical Intubation
Workplace safety: the central role of ergonomic innovations