How Artificial Intelligence can ensure a safe and healthy workplace


How Artificial Intelligence can ensure a safe and healthy workplace


Workplace safety is at the core of any modern business strategy, with working conditions being at the centre of lengthy discussions every time a worker is injured or, in pandemic times, when somebody becomes infected. Many countries and international organizations have been pushing for more and more stringent regulations through the years, to ensure that the workplace is as safe and healthy as possible. For their part, companies are quickly getting on track as well, having realized that a safe workplace is also a more productive workplace. For instance, safety improvements play a significant role in reducing workplace stress, which commonly occurs in modern companies. In fact, many workers struggle with long working hours, tensions with co-workers and managers, and general work-pressure; issues that can be alleviated by an environment that promotes health and safety, and management that actively ensures a high level of continuous monitoring.

This monitoring must be fast enough to provide help in case of a dangerous situation, and at the same time not invasive, which would lead to an even more stressful work experience. It should feel like an extra colleague is standing next to the employee always ready to help them, instead of ‘Big Brother’, constantly checking on his movements or violating their privacy. This is why Konica Minolta recently published its AI Policy, which among other things, ensures that these AI monitoring tools are properly regulated and do not jeopardise people’s privacy.


It is not a secret that the COVID-19 pandemic is one of the biggest challenges that companies have ever faced, with huge consequences for the way we look at safety and health in the workplace. As many businesses reopen and employees return to work, both workers and managers have had to face new and stringent rules to reduce the risk of infection, such as social-distancing regulations, frequent hand and equipment sanitizing, and having to wear a facemask. These regulations not only interfere with the tasks that employees must deal with but can also contribute to creating a more stressful environment.

The pandemic is not the only problem, it is just the tip of the iceberg. Safety issues have always been at the heart of the manufacturing and logistics industry, with both small and big players investing more and more in the development of efficient and better suited equipment to minimize the chance of accidents and injuries. This puts a significant weight on the shoulders of the safety managers, who every day have to make sure that safety measures are respected but have no way to continuously check if social distancing is being respected at all times or if the workers are wearing their personal protective equipment (PPE) properly. Furthermore, without automated technical support, it is impossible for them to monitor lone workers.

It must also be considered that the pandemic forced many companies to shut down for a lengthy period of time. In many cases, this led to major reductions in workforces, that also generated the need for new ways to perform the tasks that were previously executed by human workers. Here is where AI-based solutions come into play, particularly thanks to its adaptability, which makes it possible to automate certain tasks, like temperature scanning and equipment and distance monitoring.


Nowadays, thanks to the combination of IoT cameras and AI, dangerous situations and behaviours can be accurately detected, while ensuring that workers are wearing the right equipment and taking the necessary precautions for the task at hand. For instance, in those cases where workers are required to be alone for several hours near heavy machinery, the AI-based solution makes it possible to detect their activity by recognizing people who wave their hands for help, and also detect specific health-related symptoms such as chest pain or back pain, sending an automatic alarm message to the security guard.

We got in touch with Alfonso Fraire, Technical Product Owner for the Occupational Health and Safety project at Konica Minolta R&D – Digital Services, to better understand how these machine learning technologies work, particularly the activity recognition, “First of all, the algorithm detects the person. Then, it proceeds to detect the workers joints position, such as knees and elbows, drawing a sort of a stylized skeleton. Then the AI compares a sequence of stylized skeletons to recognize the worker’s movements”. He then added, “The same solutions could be also applied to the ergonomics field. For instance, we could monitor people’s movements, detecting if potentially harmful motions are carried out for long periods of time. We then can compile a report that the company can use to avoid potential long-term injuries.”

Two further use cases that are becoming more and more popular because of the pandemic are equipment control and distancing detection; both of them are significantly easier to be dealt with when the monitoring system is equipped with an AI-powered camera able to detect distances between different workers and to report any variances or violations.

  • The first use case precisely recognizes safety equipment such as masks, gloves, safety vests and glasses, making sure that they are properly worn by the worker to maximize their effectiveness and to comply with any current safety regulations.
  • The second use case detects if the required social distances are being maintained in the workplace, while at the same time monitoring safety-critical distances between people and dangerous machinery.

The important thing to remember is that the keyword for anything AI-powered is adaptability. “Our algorithm can be easily customized to fulfil the customer’s needs. For instance, we are working with a company that specialises in the production of semiconductors, which requires employees to wear a full body work suit. We’re currently working to customize the algorithm in order to detect if the suits are open, if any head-hair is protruding from the hood or if the worker’s nose is not properly fitted under the facemask,” described Alfonso Fraire, when talking about the equipment control section of the Occupational Health and Safety project.


This project area includes several solutions based on computer vision and machine learning, developed to make the workplace safer and with more intelligent automation, covering practically all the possible scenarios that might occur in manufacturing, logistic and construction companies. Some of them are:

  • Lone workers safety, which is able to recognise human activities and understand if a worker is showing symptoms of pain that could indicate illness or injury;
  • Equipment control, which detects if personal protection equipment is being worn properly or worn at all by workers;
  • Distance monitoring, which detects if the required social distancing between workers or the safety-critical distancing between workers and machinery is being respected;
  • Person detection in dangerous situations, that allows the user to create virtual perimeters into which workers are not allowed to access for safety reasons, and to detect if people enter such restricted areas. This solution can also support emergency evacuation events by detecting and notifying if anyone is still in an area that should be evacuated;
  • Temperature screening, which enables a quick and hassle-free way to monitor people’s temperature, and is applicable in all kinds of environments such as airports, offices and hospitals;
  • Cough and sneeze alert, which identifies coughs and sneezes, also ranking them based on the intensity of the sound produced;
  • Acoustic surveillance, which automatically detects acoustic events such as alarms, workers calling for help or other noises that differ from the background noise in the particular workplace or warehouse.

All these solutions are highly adaptable and are easily operated by users thanks to our web-based control dashboard, which sports a modern and easy to use UI, and a variety of charts and graphs that provide useful statistics on the alarms that were detected during a specified period.

To find out how these solutions could apply to your industry, visit the OHS webpage or get in touch with our researchers today.