Walk through a busy street and it isn’t uncommon to find most people sporting buds in their ears with their heads bowed down towards their smartphones. While they can be reliable companions when out and about, it’s this persistent distractedness that has caused a large number of pedestrians to get into accidents that could have been otherwise avoided.
For one, pedestrians who wear headphones become unaware of auditory cues, such as horns, shouts, or the sound of approaching cars. This has resulted in a large number of accidents and injuries, with figures in the U.S. tripling over the past seven years. In fact, 2018 marked the highest number of pedestrian deaths in the U.S. since 1990.
To address this, a new intelligent headphone system capable of warning users of danger in their immediate surroundings is currently being developed by a group of researchers from the Data Science Institute, Columbia.
The intelligent headphones are equipped with miniature microphones and intelligent signal processing capable of detecting sounds of oncoming vehicles. When a hazard is detected, it automatically sends an audio alert to the headphones. According to the team, this could potentially minimize pedestrian deaths and injuries once it is fully developed.
"We hope that once refined," says Fred Jiang, a Data Science Institute member and an assistant professor of electrical engineering at Columbia Engineering, "the technology will be commercialized and mass produced in a way that will help cities reduce pedestrian fatalities."
The team added how tedious and complex the research and development process was before they were able to produce a fully-functional prototype. Besides embedding a number of miniature microphones in the headset, they also included a low-power data pipeline that processes the sounds collected by the device and sends the correct cues.
The pipeline developed by the group included an ultra-low power custom integrated circuit. This is what enables it to successfully collect significant sounds with minimal battery usage. Data science techniques were also applied in designing the device. Coupled with the machine-learning capabilities of a smartphone, it should be able to quickly classify through hundreds of audio cues in the vicinity.
The researchers are currently testing prototypes on the streets of New York—a heavily noise-congested city. Jiang expressed that they will continue to develop prototypes of the system and will soon be showcasing the technology to commercial companies.
The project has already received a $1.2-million grant from the National Science Foundation in 2017. Two conference papers have already been published detailing the research in IEEE Internet of Things Journal.
For more information, visit http://icsl.ee.columbia.edu/projects/seus/.
For more news and stories, check how this digital ceiling makes smart buildings even smarter out here.