Testing real driverless cars in a virtual environment

Researchers at The Ohio State University have developed new software to aid in the development, evaluation and demonstration of safer autonomous or driverless vehicles.

This is called the vehicle-in-virtual-environment (VVE) method, which allows testing driverless cars in a completely safe environment, said Bilin Aksun-Guvenc, co-author of the study and professor of mechanical and aerospace engineering at Ohio State. allows for. ,

Imagine a driverless car parked in the middle of an empty parking lot. Although it’s driving, it’s not reacting to the real world, but to inputs from software, which tells the car what the road looks like, and what cars, pedestrians, and hazards it should expect along the way. facing.

“With our software, we enable the vehicle to think it is driving on real roads, while in fact it is driving on a large open, secure test area,” said Aksun-Guvenc. “This capability saves time, money and there is no risk of fatal traffic accidents.”

The study, recently published in the journal sensorfound that by immersing self-driving machines in a virtual environment, the technology could help the car avoid potential car collisions, enhance pedestrian safety, and react to rare or extreme traffic incidents.

Although autonomous driving technologies have become much more common on the road over the past few years, the way these technologies are tested deserves closer scrutiny, due to the sheer number of accidents caused by these systems. , said Aksun-Güvenk.

“Our future depends on being able to trust any and all road vehicles in terms of our safety, so all of our research concepts are related to working towards that goal,” said Aksun-Guvenc, who is Ohio State’s is also co-director of the Automated Driving Lab, said. A research group originally formed in 2014 to advance autonomous vehicle technologies.

Current methods of demonstrating autonomous vehicle functions involve testing the software and technology first in simulations and then on public roads. Yet this method essentially turns other road users into involuntary participants in these driving experiments, Aksun-Güvenc said, and such risks making the entire development process costly, inefficient, and potentially dangerous for both drivers and pedestrians. can be made unsafe.

To overcome the limitations of these imprecise evaluations, the researchers in this study replaced the output of high-resolution sensors in a real vehicle with simulated data to simulate its controls from a highly realistic 3D environment, such as a VR machine. headset or virtual reality glasses. After feeding data to the autonomous driving system’s computers and syncing the car’s actual motion with the simulation, the researchers were able to show that it behaved as if the virtual environment were its real surroundings in real time.

But Levent Guvenc, co-author of the study and co-director of the Automated Driving Lab, said that what makes their software particularly powerful is how flexible their virtual environment can be. “When real senses are replaced by virtual senses, the model can be easily changed to fit any type of scenario,” Guvens said.

Because the VVE method can be calibrated to maintain real-world properties while modeling rare events in a virtual environment, it can easily simulate extreme traffic scenarios, such as someone jumping in front of a vehicle. For mundane people like pedestrians waiting at crosswalks, he said.

Additionally, with the help of a communication app for vehicle-to-pedestrian connectivity, the software can use Bluetooth to communicate between a pedestrian with a mobile phone and a phone in the test vehicle. The researchers actually had a pedestrian speeding down a simulated road at a safe distance from the test vehicle. But the Bluetooth signal told the car that the person was driving right in front of it.

“The beauty of this method is that road users can share the same environment at the same time without having to be in the same place,” Guvenc said. And although it can take time to create these super-realistic environments, he said the technical challenge of coordinating different environments to use in real-time simulations is one his team has solved.

The team has also filed a patent for the technology. Guvens said in the future he would like to see it integrated into traffic guidelines created by groups such as the National Highway Traffic Safety Administration.

“We could see this technology becoming dominant in the industry in the next five or 10 years,” Guvenk said. “So we’re focused on building more applications for it.”