Ground truth is the ‘guiding light’ for ADAS and AD progress

Vehicles are growing smarter and safer all the time. More sensors, advanced artificial intelligence (AI) and greater computing power promise to revolutionise the safety and efficiency of roadways, but just how safe is safe? Boston-based NextDroid offers a software platform that collects and analyses vehicle and road data to help developers validate the performance of advanced driving assistance systems (ADAS) and automated driving (AD) features.

It consists of a roof-mounted hardware sensor SLED and a cloud-based software toolkit that together analyse test-drive data. Developers can fix the Veritas platform to a test vehicle—on the roof and in the trunk, with software throughout—to identify driver and AI error. The detailed feedback can not only help engineers improve system performance at a much more granular level but also serve as an objective measure of safety.

Measurable safety

“Traditionally, the automotive industry has focused on accidents,” says David Bruemmer, Chief Strategy Officer at NextDroid. “But why don’t we instead look at safety prior to the impact of the vehicles colliding? Let’s understand what happened: did these two vehicles come within one foot or one centimetre of each other? Without objective ground truth, we simply don’t know.” The NextDroid team are taking a much deeper dive into the system performance.