Real-Time Data and A.I.: How Collision Avoidance Taps New Tech to Boost Facility Safety

Data — it’s a word that resonates in today’s world. With information flowing from every corner, we find ourselves surrounded by countless data points: think about how many times an operator starts and stops a fork truck and how often a fan in the facility turns on and off. The availability of data seems endless. However, amidst this abundance lies a challenge that many of us face — how do we sift through this data overload, avoiding paralysis and inaction, to decipher useful information?

Real-Time Data Can Drive Actionable Insights

What does actionable data look like and how can we collect it? Focusing on collision avoidance, we can start by identifying zone breaches around a piece of equipment, which happen when an operating vehicle gets too close to someone or something. We can also consider breaches found in stationary applications like intersections.

With zone breaches, the initial step is getting the necessary information. Vision-based A.I. collision avoidance technology provides knowledge of the incidents occurring, and its camera system offers the added benefit of having photos of the incidents that can tell a story. These photos aren’t biased, emotional, or open to interpretation — they provide a true picture of what’s occurring.

As we start to leverage the available data, we can begin to move toward actionable insights, such as:

• At this intersection at this time of day, the data sees a higher number of interactions.
• We have four different brands or types of lifts in our facility, and the data shows significantly more incidents on one type or brand in a specified area of the operation.
• There are more close calls in one area than anywhere else.

These data points also drive near-miss initiatives, providing a huge opportunity to prevent accidents, injuries, and material damage — helping move toward a safer operation.

Visual Artificial Intelligence – Data + Photo Documentation

In simple terms, vision-based A.I. involves taking thousands of photos of an object or group of objects we want to detect and training a model on those photos. A machine learning algorithm then determines, based on the objects you’ve taught it, a confidence level for any object it identifies in a new frame. Over time, an object library is built, allowing operations to decide, in the case of collision avoidance, which objects should trigger alerts.

During the past five years, Matrix has been dedicated to updating and refining the model, minimizing false alerts and optimizing processing speed. An entire analytics team works to filter and decipher the actionable data.

OmniPro Collision Avoidance System Taps the Best of Real-Time Data and A.I.

Through this research, Matrix has developed the innovative OmniPro collision avoidance system. OmniPro enables line-of-travel, crosswalk, and blind-spot pedestrian and vehicle alerting for mobile equipment. This award-winning* system works without personal wearable devices or tags. It not only “sees” and identifies people and hazards, issuing visual and/or audible alerts, but also photograp*