PAWS Headphones Warn Wearers of Oncoming Traffic Using Audible Alerts

Public transportation users, joggers/walkers, and audiophiles could benefit from Data Science Institute’s (Columbia University) PAWS (Pedestrian Audio Warning System) headphones, which alerts wearers of oncoming traffic to avoid accidents. We’ve all seen the videos of people walking while texting, or using other smartphone apps, only to fall into a pool or walk into walls and other obstacles. While most of those videos are funny, not paying attention to your surroundings can have harmful consequences.

While they don’t look or perform like Beats By Dre, the PAWS headphones do pack a ton of technology, and could one day save your life. To detect oncoming traffic, the headphones are outfitted with five MEMS-based microphones positioned at each ear, back of the head, and on the chest of the user, which allows the headset to pick up vehicles in any direction. A custom PCB outfitted with an STM32F446RE MCU grabs the analog signals from the microphones and extracts the acoustic data, which is then processed using a machine learning application on a Bluetooth connected smartphone.

The machine learning classifiers analyze hundreds of acoustic signatures from city streets and nearby vehicles, separating those (people conversing, birds, hotdog vendors, etc.) that are inconsequential for safety. While the PAWS headphones are great at identifying oncoming traffic, it can’t detect a vehicle’s trajectory, much less multiple vehicle trajectories. That said, the researchers are looking at ways to improve the platform, which could be beneficial for construction workers repairing highways, or police officers directing traffic.