Bird's Pyramid: A Fresh Look at Prevention in Construction

Bird's famous pyramid reminds us of a simple truth: serious events are often preceded by a multitude of more discreet signals, near misses, and a number of situations that could have gone wrong.

During my years in the field, I've seen this pyramid at work, sometimes blatantly, sometimes more insidiously. We often focus, and rightly so, on High Potential Events (HPEs) to draw crucial lessons from them. But let's not forget that these HPEs are rarely isolated cases. They are often the culmination of an accumulation of small things, those near misses that, when added together, create a riskier environment.

Today, with the evolution of our professions and the increasing complexity of construction sites, it can seem difficult to capture all these weak signals. Teams are sometimes dispersed, prevention officers must juggle between several sites, and information is less easily passed on. We might then be tempted to think that tracking down the base of the pyramid is too daunting a task, that it's better to focus on what's visible, on what has already almost gone wrong.

Yet, this is precisely where artificial intelligence (AI) and computer vision offer a new and promising perspective. Imagine for a moment being able to have eyes everywhere on your construction sites, without multiplying the number of prevention teams. This is what innovative solutions like those developed by Cad42 make possible. For a very reasonable investment—a few hundred euros per month—these technologies continuously analyze the images from your cameras and can automatically identify risky situations that previously went unnoticed.

Let's take some concrete examples, those that we unfortunately come across all too often: a Light Individual Rolling Platform (PIRL) used without its stabilizers being correctly deployed, a pedestrian path crossed by a machine moving in reverse, a hopper left without adequate protection, a missing temporary guardrail, or a textile sling that shows signs of deterioration. These are not necessarily intentional "breaches of the rules", but rather moments of inattention, habits adopted without necessarily measuring the danger, or a one-off lack of vigilance.

AI, on the other hand, doesn't judge. It observes and alerts. It allows us to collect an impressive amount of objective data on situations that could have gone wrong. By analyzing this data, we gain a much more accurate view of the reality on the ground and can thus measure the maturity of our prevention culture. Are we in a phase where near misses are frequent? Where certain types of risky situations recur regularly?

With this factual information, construction companies can adapt their prevention actions in a much more relevant manner. Instead of general awareness campaigns, they can target the most common risks, strengthen training on the critical points observed, and adjust procedures to make them even clearer and more applicable in the field.

AI and computer vision aren't there to replace humans—far from it. They're a powerful tool to help us, construction professionals, gain a more comprehensive and objective view of what's really happening on our construction sites. By reinvesting in Bird's base-of-the-pyramid analysis using these technologies, we're not just reducing the risk of serious accidents. We're building a more mature culture of prevention, where collective vigilance and the correction of risky situations become natural reflexes for everyone's safety. This is a positive step forward for our sector, and I'm convinced it will help us make our construction sites ever safer.

Tribune by Igor CANONNE, HSE 4.0 Solution Manager, CAD.42 (LinkedIn).