Aptiv to share its self-driving car data to help other researchers
The dataset includes 1.4 million images from Boston and Singapore.
Jake HolmesReviews Editor
While studying traditional news journalism in college, Jake realized he was smitten by all things automotive and wound up with an internship at Car and Driver. That led to a career writing news, review and feature stories about all things automotive at Automobile Magazine, most recently at Motor1. When he's not driving, fixing or talking about cars, he's most often found on a bicycle.
Automotive technology company Aptiv wants to help advanced research into computer vision and autonomous-driving technologies by sharing some of the information it has already gathered in the real world. Aptiv announced this week it will share what it's calling the largest public dataset to date of autonomous driving data. The data is open-source and free to use.
The data-sharing project is called nuScenes and consists of data that Aptiv gathered in the real world on roads in Boston and Singapore, meaning both left- and right-hand-drive traffic is covered. nuTonomy, an Aptiv company, has operated self-driving test cars in both Singapore and Boston, which is most likely where the data comes from.
The database consists of 1,000 scenes each 20 seconds long, with a total of 1.4 million images, 390,000 lidar scanner outputs and 1.4 million human-annotated objects. Aptiv says it gathered the data from six cameras, five radar units and one lidar unit on each car, giving a 360-degree field of view around the vehicle.
The goal is to help other researchers evaluate safety and develop software for self-driving vehicles. Aptiv says that 1,000 individuals and 200 academic institutions have already signed up to access the data. Previously, Aptiv says most of this type of data was kept private by research groups and companies.