Scientists at Ohio State University developed new software for training and testing unmanned vehicles. A technique called “Vehicle in a virtual environment” (Vehicle-in-Virtual-Environment, VVE) allows you to test autonomous vehicles in a completely safe environment, saving time and money, and eliminating the risk of accidents.

The program creates a virtual environment in which an unmanned vehicle is immersed. It reacts to virtual roads, cars, pedestrians and obstacles as if they were real, thereby learning how to drive safely in a variety of situations, from ordinary to extreme.

During testing, the car can drive on the test site, thinking that it is moving on a real road. Moving around the polygon, the car does not react to the real world, but to the input from the software that generates objects on the way.

Simulation of the real world in the car

The study shows that such immersion can improve pedestrian safety and the vehicle’s ability to respond to rare or extreme traffic events. Unlike current approaches that test autonomous technologies first in simulations and then on public roads with other road users, VVE is safer and more efficient.

Test equipment

In addition, the program includes a Bluetooth capability for communication between the pedestrian and the car. The scientists quickly ran across the simulated road at a safe distance from the test vehicle, and a Bluetooth signal indicated to the vehicle that a pedestrian was indeed crossing in front of the vehicle, alerting the vehicle to a possible collision.

Scientists have applied for a patent on this technology and suggest that it could become the industry standard within the next 5-10 years.

We previously reported that the Privacy4Cars V released a new tool , which shows what information about a particular vehicle can be collected. By entering the vehicle’s VIN, the user can see what automakers might know. It is estimated that cars can produce 25 GB of data per hour.