Recognizing the sights and sounds of emergency vehicles

Whether it’s the high shrills of a fire truck or the flashing lights of a police cruiser, it’s important for every driver to recognize the telltale signs of an emergency vehicle. Over the last few years, we’ve been teaching our self-driving cars to detect and respond to everything from fire trucks and ambulances to police cars and police motorcycles. Last month, we kicked off our latest efforts to improve the way we detect emergency vehicles.

In collaboration with Chandler Police and Fire, we set up the first emergency vehicle testing day with our new self-driving minivans. Throughout the day and into the night, our vehicles — equipped with a powerful suite of sensors — observed police cars, motorcycles, ambulances, firetrucks, and even a few undercover vehicles as they trailed, passed and led our self-driving cars. As these emergency vehicles drove around, our sensors collected samples at various speeds, distances, and angles — all with the goal of building up a library of sights and sounds to help teach our self-driving cars to respond safely to emergency vehicles on the road.

Today, our self-driving minivans are capable of hearing twice as far as they could with our last suite of sensors. And with our new sensors, we’re teaching our software to discern which direction the sirens are likely coming from. Drivers often hear the sirens of an emergency vehicle first, so being able to gauge the direction of an oncoming fire truck or ambulance allows our self-driving vehicles to make smarter and safer decisions. For example, our vehicles may pull over if an emergency vehicle is coming from behind, but yield at an intersection if our sensors detect sirens coming up ahead.

All this is possible because our self-driving minivans are outfitted with custom-built self-driving sensors, including an audio detection system designed in-house. Not only have we improved the way our cars hear sirens, but our vehicles can see emergency vehicles and their flashing lights even further and clearer with our custom vision system, radars, and LiDARs. These advanced sensors allow us to both reliably detect emergency vehicles in real-time, and collect better information to train our software. That way, when we’re on the road, we can recognize an emergency vehicle even if we’ve never encountered it before (or if it appears differently because of the angle or lighting conditions).

One of our self-driving cars pausing at an intersection to let an ambulance pass, even as our light turns green and other vehicles begin moving

This training is key to reliably detecting and responding to emergency vehicles in Arizona and beyond. By teaching our cars this advanced capability, we’re moving closer to bringing truly self-driving technology into the world.