Ascent Robotics has announced it has raised $10m to fund research into its driverless cars AI tech, used to ensure the vehicles can operate safely on the roads.
The company has built a self-driving car virtual simulation platform to help inform self-driving cars how to operate safely.
Subjects sit in front of a screen and steering wheel, with a VR headset on. Various obstacles and scenarios are thrown at them and they must react exactly as if they were driving a car. Ascent Robotics believes this is the best way for autonomous cars to learn how to deal with certain scenarios.
Although the majority of businesses experimenting with driverless cars use real vehicles on the road to see how they operate and build their intelligence, Ascent Robotics believes using VR is a safer option.
Also, only a small portion of the data on real roads can be used to understand the intelligence needed to drive safely on the road, in all situations. Using VR means the company can throw all kinds of hazards in the cars’ way, in a safe environment and learn how those controlling the cars react.
“Driving data isn’t everything,” Ascent Robotics’ founder Fred Almeida told Bloomberg. “We can learn key behaviors in a simulation that transfer effectively well to a physical system.”
The simulation environment is built upon Unreal Engine – the same platform used to create games such as Batman: Arkham Knight and Assassin’s Creed Chronicles. It features pedestrians, buildings and traffic signs, but the company explained it’s not a real city, but a completely new, made up one. The blocks can be changed to suit the day’s particular exercise.
“You want to have that uncertainty and noisiness of human driving as part of the learning experience of the car,” Almeida added. “This should result in a more robust result at the end.”