Samsara, the pioneer of the Connected Operations® Cloud, today launched its Drowsiness Detection feature into general availability for customers globally. Drowsiness Detection uses Samsara’s comprehensive AI models—trained on its large-scale dataset—to proactively detect when signs of drowsiness occur. It then triggers real-time in-cab audio alerts for drivers, while notifying managers via text or email to triage fatigue-related events in the moment.
Image: Samsara
These insights can be viewed as aggregated reports within the Samsara Platform, allowing managers to analyse patterns of fatigue across their fleet, focus on driver coaching, and ultimately reduce drowsy driving to improve road safety.
According to the European Commission’s Safety Performance Indicator, driver fatigue is a contributing factor in 15% to 20% of crashes. In the commercial fleet industry, drivers are even more at risk given long hours and unpredictable road conditions. That’s why preventing drowsiness is a top priority for Samsara’s tens of thousands of customers. While advancements in AI and machine learning have made proactive alerts possible, drowsiness remains an incredibly nuanced behaviour to train AI models to detect.
“It’s hard to detect when someone is truly drowsy. It’s more than a single behaviour, like yawning or having your eyes closed. Drowsiness can be less common than other risky driving behaviours, so accurate detection is only as good as the data that feeds and trains AI models,” explained Evan Welbourne, VP of AI and Data at Samsara. “That’s where the scale of Samsara’s data sets our solution apart. We train our models on more than 38 billion minutes of video footage within our platform to provide high accuracy and impact for our customers.”
Image: Samsara
To ensure accuracy, Samsara’s comprehensive Drowsiness Detection is trained to consider several behaviours that indicate fatigue, in alignment with leading and clinically validated standards for defining drowsiness. These behaviours include: head nodding, slouching, prolonged eye closure, yawning, rubbing eyes, and more. Yawning alone is often not a sufficient detector of drowsiness. In fact, an analysis among early adopters of Samsara’s Drowsiness Detection found that approximately 77%1 of drowsy driving events were detected by behaviours other than yawning alone.
Samsara first announced Drowsiness Detection at its annual Beyond conference in June, which gathered more than 2,000 physical operations leaders across the industry. Since then, customers have already experienced meaningful impact. For example, a large oilfield services company has seen a significant decrease in how often drivers are falling asleep during shifts since using Samsara’s Drowsiness Detection.
Samsara’s petabyte-scale dataset collects more than 10 trillion data points each year and is used to train AI models that automate workflows, accelerate time to value, and provide personalised, actionable insights for customers. In one year alone, Samsara’s Video-Based Safety solutions have helped prevent more than 200,000 crashes among customers worldwide.
