With an aging population in need of transportation, Japan is betting on self-driving cars — © AFP Kazuhiro NOGI
Autonomous driving technology can potentially reduce traffic accidents. However, the more sophisticated automotive computers become, the greater the risk of cyberattacks for autonomous vehicles.
Similar cyberattacks can endanger not only those of the affected vehicle, but also those of nearby vehicles and buildings. Road testing to improve safety is expensive and time-consuming, but researchers at the Commonwealth Cyber Initiative (CCI) developed a virtual option: a vast digital library of real-world driving scenarios to help manufacturers design safer autonomous vehicles.
A team from the Virginia Tech affiliate Global Center for Automotive Performance Simulation (GCAPS) is working with researchers from the Virginia Tech Transportation Institute (VTTI) to validate the safety and performance of automated systems technologies virtually. His project is supported by the Innovation: Ideation to Commercialization program of the Commonwealth Cyber Initiative in southwest Virginia, which provides funding to commercialize technology generated from research at the intersection of data, security, and autonomy.
Scenario-based virtual testing is a safe and cost-effective alternative to traditional distance-based validation of autonomous technology, which requires the onerous and sometimes impossible task of driving billions of miles on public roads.
Miguel Pérez, an associate professor of biomedical and mechanical engineering who directs the VTTI Data Engineering Program.
Pérez explains in correspondence to Digital magazine: “The library establishes the truth on the ground based on real-world data for vehicle interactions with the road, infrastructure, vulnerable road users and other vehicles,” said Miguel Perez, associate professor of biomedical engineering and mechanic who leads VTTI’s data engineering program. .”
lots of real data
According to Pérez, to “build a highly reliable virtual training library, the research team is tapping into a unique resource: the vast trove of real-world events recorded in VTTI’s naturalistic driving databases.”
He further explains his company’s innovations: “For more than 30 years, VTTI has been collecting data from cameras and sensors installed discreetly (with permission) in ordinary cars belonging to ordinary people across the country. The data, which includes basic vehicle dynamics, video, and sensor information, represents more than 70 million miles of real-world driving behaviors exhibited during normal daily commutes, culminating in more than 10 petabytes so far. ”.
With recent innovations at his company, Perez notes, “For the past five years, GCAPS has worked with VTTI to apply data to automated driving technology. The team has been developing an algorithm to classify the lateral and longitudinal micro-movements of a vehicle by assigning objective values to the displacement of a vehicle at given speeds and distances. Once an event has been classified, it can be recreated: turned into a simulation-ready dataset that includes trajectories, positions, road features, and terrain.”
Researchers have been exploring questions like how close people actually drive to each other. Does a person’s age affect their reaction? Would a person in a different region react differently?
In case of a cyber attack
With the cyber security focus, the set of questions has been expanded to include what might happen to vehicle performance in the event of a cyber attack. How will a vehicle compensate for a counterfeit sensor? Will the autonomous vehicle be safe on the road if it is actively being hacked?
In answering this, Pérez states: “Because the interactions in the events represent real-world data, the precision and timing of the interactions provide a very valid basis,” Pérez said. “As such, the library makes it possible to determine the severity of the cybersecurity risk associated with vehicles, but in a safe and repeatable way through simulation.”
This capability will help engineers determine requirements for automated vehicle cybersecurity products, such as response delay times required to minimize vehicle issues or the percentage of computing resources that can be safely reallocated to defend a hacked network.
