Vehicle-to-vehicle communication technology (V2V) looks like a promising means of reducing car accidents, but it faces one big hurdle: Cars don't currently carry the transceivers required to make it work. Enter Nexar, an app that turns your smartphone into a V2V device.
On first blush, the Nexar app looks like a dashcam, a device that records everything in front of your car, and automatically saves a portion of video and telemetry in case of an accident. Behind the scenes, however, the Nexar company runs a deep learning platform, analyzing data from its apps on the road to look for potential collisions.
Most accidents happen when one driver makes a move that other drivers don't see, often because we are only capable of looking in one direction at a time. Automakers and the government propose that V2V can go a long way to eliminating these sorts of accidents, as this technology makes each car aware of the locations and trajectories of all other cars in the vicinity. When two cars approach an intersection, broadcasting their speeds and directions, they can each alert their drivers, even if those drivers can't see the other car due to a visual obstruction.
To get a sense of how the Nexar app can give current cars V2V technology, I went out on a test ride with Amit Shalev, Nexar's West Coast Growth & Operations Lead. Our car had a phone running the Nexar app mounted to the windshield, and another car driven by a Nexar staffer, also equipped with the app, took the lead to simulate an incident.
The running app showed the video footage it gathered as we drove, using its camera to see everything in front of the car. Shavel told me that the footage and other sensor data from the phone, such as its GPS location and accelerometer information, would be uploaded to Nexar's cloud and analyzed with its deep learning stack.
With deep learning, a computer can identify specific objects in an image. Nexar is training its stack to recognize school buses and garbage trucks, among other types of cars, along with roadside infrastructure and pretty much anything else in the driving environment.
As we drove down the freeway, Nexar's other car, much farther ahead in our lane, slammed on its brakes. The Nexar app on our windshield sounded off with a beep and showed a warning graphic, including the time until we would hit the car ahead. Given the warning, we were able to brake to avoid a collision.
Using Nexar's V2V capabilities, we would have gotten the warning even if the car doing the braking was a couple of cars ahead of us, where we wouldn't be able to see it.
Behind the scenes, the app in our car sent its telemetry information to Nexar's cloud, while the app in the car ahead did the same. It used the camera data to confirm our lane position, bolstering the GPS location. When the driver ahead hit the brakes, the phone's accelerometer sensed the change in velocity, and the app sent that information to the cloud. Nexar's back-end platform computed that our car, if it remained on the same path, would collide with the braking car ahead, and so made the app issue its warning.
Beyond this scenario, Shalev explained that Nexar could be put to many other uses. For example, the app is currently in use by the Via car service in New York City. When the app recognizes a garbage truck on an avenue, it automatically alerts the other Nexar apps being used by Via drivers on that route, so they know to avoid the lane in which the truck is driving.
Of course, like any crowd-sourced service, Nexar thrives on distribution. If the car braking ahead of you isn't running the app, you won't get the V2V warning. However, if the car is close enough and within the camera's view, the app itself can run a quick calculation and determine if you are in danger of a collision, working much like the precollision warning systems automakers are beginning to install in cars.
Along with this built-in precollision warning, the Nexar app also works like a dashcam, saving 20 seconds of video before and after an incident.
Shalev told me that the cities with the greatest density of Nexar-using drivers are New York, Tel Aviv and San Francisco. The company is partnering with fleets to get more of its app out on the road, collecting data. He pointed out that if the company captures just one to four percent of an area's driving hours, that would create a very robust data set for its deep learning platform.
Running the Nexar app while you drive will certainly use up the phone's battery, so be prepared to charge it when you get to your destination. And so drivers can use navigation while using Nexar, the company integrated the app with Waze, so you can still hear turn-by-turn directions.
When I asked about data usage, Shalev said that the app transmits a minimal amount of telemetry data while underway. It does, however, store video from driving, uploading it to Nexar's cloud when the phone is on a wifi connection. That video data lets the deep learning platform analyze driving information, so it can get smarter about recognize everything in the environment.
To be honest, I don't see too many drivers popping their phones to a mount on the windshield whenever they set out on the daily commute or a shopping trip, despite the safety implications. But Nexar seems primed to work as the backend for new connected cars. For example, if GM began to send and receive telemetry data from its cars, which already come equipped with a 4G data connection through Onstar, to Nexar, it could implement V2V almost overnight.
Along with data connections, many new cars are getting forward-facing cameras, which could be leveraged to feed video to Nexar. And cars already have a wealth of sensor data from GPS location to speed, to more esoteric bits such as whether the windshield wipers are on and if the traction control gets engaged.
Crowd-sourcing all these bits of data could make for much safer driving, and Nexar could work as the glue that ties them together.
For now, however, we will have to be satisfied with the Nexar app. Available for Android and iOS, the Nexar app is absolutely free, as the company is turning to fleet installs for revenue. And to make things easier, Nexar offers a phone mount for only $6.95 on Amazon.