Nexar Dash Cams help in training autonomous cars

Tel Aviv-based Nexar reported an increase of 30% in the usage of their smart car cameras (Dash Cams). As of today, 210 million kilometers are being filmed and recorded monthly, and a total of 3.2 trillion road pictures have been gathered.  The company also estimates that the scope of its cameras usage will double in the next 12 months.

Nexar was founded in 2015 by CEO Eran Shir and CTO Bruno Fernandez-Ruiz, and currently employs about 100 people – about 80 in Israel and the others in New York.

The company developed GPS-based dash cams, connected to the Smartphone through designated app, which allows recording and providing insights regarding the road users. The app takes advantage of the sensors and processing capabilities of the Smartphone in order to provide safety alerts to the driver, such as deflection from the driving lane or danger of collision.

Main feature of the app is the transmission of information to the cloud, which in turn, through V2V (Vehicle-to-Vehicle) network, it provides alerts to other drivers around, alerting them of dangers beyond their field of vision that they can’t directly notice.

At the same time, the rapid development of smart mobility and autonomous driving are expanding the target markets for the data collected by the camera. “The data collected enables automakers to train the cars they manufacture and keep their maps up-to-date in real time to accommodate for road changes”, said CEO Shir. Currently, the company provides information to companies which develop autonomous driving systems, regarding edge cases of driving scenarios, in order to train their systems and meet the 4 & 5 level requirements. This area is crucial – but lacks the sufficient data. All this leads to a new product launched by the company six months ago: a service which provides drivers and insurance companies with a detailed reconstruction of the actions of all parties involved in a collision. This service is provided using raw camera data, on-board sensors, and deep learning AI algorithms. 

Using this method, data is collected from the dash cam and several sensors, integrated in accurate road maps, and reconstructs the collision developments step by step, including car behavior and the hit moments. Currently, Nexar started to provide this service to hundreds of thousands of the Japanese insurance company Mitsui Sumitomo Insurance customers.

Another new service provided by Nexar, is a highly detailed road maps for the autonomous car manufacturers. Today, most dynamic road maps are produced through LiDAR technology, which support relatively low detailing level. The company announced lately that its dash cams provide up-to-date details that are easy to incorporate in existing road maps.

These up-to-date details can also be used for road maintenance. Recently, Nexar and U.S.-based Blynscy announced a partnership to collect data on 6.5 million kilometers of public roads. This data will be used to locate road damages and transfer the information to the authorities. National studies show that effective marking of roadways saves lives. Over 50 percent of fatalities on America’s roadways — roughly 19,000 deaths annually — result when motorists leave their travel lanes, according to federal data. Most pavement markings are repainted only once a year, but with harsh weather conditions and degrading asphalt pavement markings oftentimes need to be repainted or maintained more often.