KUNGFU.AI developed an AI-powered early warning system for Waste Management that detects oncoming vehicles in real time, alerting rear loader riders of danger and preventing life-threatening collisions.
WM, North America’s leading waste services provider, had a critical mission—keep their rear loader riders safe. These riders face life-threatening risks from rear-end collisions while collecting residential waste. The vision? Create an automated detection system that could serve as an extra set of eyes, warning riders of oncoming danger and ultimately saving lives.
Relying solely on driver and rider awareness to avoid collisions wasn’t enough. Waste Management needed a proactive solution—one that would detect oncoming vehicles and provide real-time alerts to riders at risk. But this wasn’t just about developing a smart detection model; it needed to work at the edge, on low-power devices, with speed and precision to prevent accidents in real time.
Waste Management partnered with KUNGFU.AI to bring their vision to life. We developed a high-performance platform using an ensemble of computer vision models running on edge devices powered by Google Cloud Platform. The device, equipped with Edge TPUs and a USB battery, combined detection and segmentation frameworks to differentiate trailing vehicles from other non-threatening objects.
The system didn’t stop at detection. We built a threat prediction model that calculated speed, velocity, and distance to assess the danger level of approaching vehicles. If a vehicle posed a threat, an embedded sensor alerted riders to brace for impact—giving them valuable seconds to react and avoid injury.
The Waste Management Early Alert Rear Safety Device was field-tested and proved its effectiveness, successfully alerting riders to potential collisions. During pre-deployment parking lot tests, detection took just 0.05 seconds per frame, with tracking and scoring also clocking in at 0.05 seconds per frame. The result? Highly accurate detection with no false negatives reported. This AI-powered solution is paving the way for safer operations and giving Waste Management’s riders the protection they deserve.
Segmentation model that can understand the difference between people, objects, and moving vehicles.
Demonstration of a model calculating the threat of an oncoming vehicle.
“The things that stood out immediately while working with the KUNGFU.AI team are the caliber of the resources that they bring in to bear and the other aspect is just how well they’ve worked with our team. KUNGFU.AI has gone out of their way to make it easy for us to work with them. These are people who are both smart and knowledgeable with different techniques, which is something we’re looking for out of our partnership.”
