• Homepage
  • >
  • Videos
  • >
  • Cognata Uses Deep Learning Tech to Gather Autonomous Driving Data [Video]

Cognata Uses Deep Learning Tech to Gather Autonomous Driving Data [Video]

How can the virtual world impact the real world? Well, when it comes to autonomous vehicles, a simulated reality could help accelerate the cars’ ability to enter the physical realm.

Self-driving cars would need about 11 billion miles in data in order to prove they are safer than human drivers, according to a published report. With a fleet of 100 cars running 24 hours a day, that would take 500 years, according to the report. But Cognata, the Israel-based company that just landed $5 million in funding from Emerge, Maniv Mobility, and Airbus Ventures, is able to gather a few million miles in a few hours, CEO Danny Atsmon told Mobility Finance.

“[If] OEMs and Tier 1 want to launch autonomous vehicle in 2020, 2025, then we are accelerating the pace of [getting] autonomous vehicles on the road,” he said, adding, “This is our goal.”

Cognata recently released a video showing just how their simulation works. The company uses a deep learning engine that recreates real cities, like San Francisco, in 3D. Every tree, sidewalk edge, and stop light is just where it is in real life, Atsmon said, adding that the simulation works by layering in three different layers of data. The first layer is the 3D model of the city, the second is dynamic elements, like the movement of cars and pedestrians, followed by the third layer of sensing technologies, like radar, sonar, lidar, and cameras.

The benefit to simulation tests, beyond just gathering millions of data points, Atsmon said, is that all kinds of difficult driving conditions can be easily rendered — like driving on a rainy night — which means engineers don’t have to wait around for inclement weather.

“It’s really hard to reproduce [that] in real life, but you can reproduce this in a simulation,” he said. “And you can teach an autonomous system how to react… Another part of the simulation is – think about how the Cognata simulation is using deep learning and artificial intelligence to create San Francisco and other full cities, with all [their] traffic and weather conditions.”

Check out the video below to see the capabilities of the Cognata Deep Learning Autonomous driving simulation engine:

  Like This Post