Training self-driving cars to make way for emergency vehicles
For self-driving cars to become commonplace, they will need to be equipped to handle every possible traffic scenario that could come their way. That includes making way for ambulances, police cars and other emergency vehicles, which is why Google spin-off Waymo has begun training its Chrysler minivans to respect the sound of wailing sirens.
Google turned its self-driving car project into its own standalone company, Waymo, in December of last year, but it actually begun investigating the emergency vehicle dilemma well before then.
Its self-driving report from November 2015 described a library of various siren sounds and the development of software that could identify them and slow down the vehicle. Last year, it filed a patent application for technology that detects an emergency vehicle's flashing lights.
And things seem to be moving along just nicely. Waymo now says its self-driving Chrysler minivans, which were incorporated into its fleet in May last year, boast a new set of sensors that can hear sirens twice as far away as the previous set. These work with LIDAR, cameras and radar to also detect the emergency vehicle's flashing lights.
Waymo is training the software to not just detect the sirens, but pinpoint the direction the sirens are coming from, so when a fire truck or ambulance enters the area, the self-driving car can make smarter decisions. If an ambulance is coming from behind, for example, the car might pull over, while if it is coming from up ahead, the car might yield at an intersection.
This technology was put through its paces at an emergency vehicle testing day, which was run in collaboration Chandler Police and Fire. During both the day and night, Waymo's Chrysler minivans used their upgraded sensors to detect police cars, motorcycles, ambulances, firetrucks and even undercover vehicles.
Waymo says that throughout the exercise, the vehicles reliably detected the emergency vehicles. All the while, it gathered data on the various speeds, distances and angles of the approaches to continue building out its library of sights and sounds, which will keep the cars out of emergency responders' way when they hit the road for real further down the track.