Benefits of AI in Hi-Vis and PPE Monitoring

Occupational health and safety professionals sometimes describe personnel protective equipment (PPE) as the ‘last resort’. This can lead workers to believe that PPE isn’t important. It is undeniably more effective if you can eliminate hazards at source: if there are no vehicles, they can’t hit a pedestrian; if there is no machinery, people can’t get a limb trapped inside. If vehicles and machinery are essential to your business, can you reduce risk as a source, for example with slower vehicles and better designed machinery?

The next most effective measure is to prevent people getting close to physical barriers. But once you have done everything practical to eliminate, reduce or prevent access to hazards, PPE is the final barrier. Describing it as the ‘last resort’ doesn’t mean that it isn’t important. The global market for PPE in 2023 is estimated to reach nearly $70 billion and there is no sign of it plateauing. 

While the best strategy to avoid objects falling on people is to reduce the likelihood of things falling and to prevent people walking underneath objects that could fall, a hard hat provides an additional layer of protection. In busy environments where pedestrians and vehicles might come in close proximity, wearing high visibility clothing increases the chance that a driver will see the pedestrian. Although PPE should be the first control a safety manager thinks of, it can make the workplace safer.

If you expect people to wear all PPE all the time, they can become blasé about the requirement. A worker might think, ‘I know I don’t need the high-vis jacket when I’m inside my vehicle, so maybe I don’t really need it when I get out’. Control what you can without PPE – and then be clear about when PPE is really needed. Make sure that PPE is available in the right sizes and specifications, when and where people need it.

Each item of PPE only protects one person, and only when correctly selected and used. The wrong gloves, a damaged hard hat, and high vis clothing tied around the waist are ineffective. 

Accurate information about when and where prescribed PPE isn’t being used will provide insights on the reasons for non-use or ineffective use. If most of the people who don’t wear PPE carry out the same manual handling task, perhaps there is something about the range of movement required by that task that makes it uncomfortable to wear PPE. You might need to change the handling task to change the PPE behaviours. If people are wearing the old hard hats that you tried to dispose of, ask them what’s wrong with the new hats, and next time build in a worker-piloting phase for new PPE.

But how do you know when people aren’t wearing PPE? People might hastily don their hard hats and high vis when they see you coming. If you spot someone without gloves, how do you know how widespread the practice is? You need to see the patterns in non-use to make useful decisions. We know the industry recognizes this need. In a survey carried out by ProtexAI in October 2022, seven out of ten people selected being able to detect when PPE is being worn as a desirable function of AI.

One technical solution being trialled is to build smart technology into the PPE. A hard hat will ‘know’ when and where it is being worn, and in some cases has added sensors which report when a worker is tired, over-heated or collapses; gloves report when they are damaged; hearing protection also measures noise exposure. While some of these products are being used successfully, their role in PPE violation detection has been limited. Barriers to adoption include the greater expense of smart PPE, and concerns over how bio-data (such as heart rate) will be used.

An alternative technology solution is PPE detection using computer vision (CV). CV can make use of existing CCTV networks, and can be trained to recognize the types of PPE that you want people to use. Your AI PPE monitoring system could check that workers wear hard hats when within a specified distance of an overhead hazard, or respiratory protection where dust can’t be controlled by other means. 

While CV will provide the data, the insights and the decision-making is in your hands. For example, the data shows you that non-compliance with glove wearing increased when you recruited new workers, and reduced again after training. This gives you the evidence to make a business case for providing a safety induction which includes instructions about PPE on the first day. You might detect changes in compliance when you change glove supplier, and can quickly identify problems with the new product – before people start reporting skin problems from exposure.