The data was collected over the span of one year, and it comes from multiple self-driving research vehicles. It includes LiDAR and camera sensor data, GPS and trajectory information, as well as 3D point cloud and ground reflectivity maps. Ford is even sharing a plug-in with data visualization tools.

Ford

While other companies, like Waymo, have shared self-driving vehicle data in the past, the breadth and depth of Ford’s data is unusual. Because it was collected over an entire year, it includes a variety of weather conditions, including rain, sun, clouds and snow. The data was gathered in the Metro Detroit area, so the vehicles experienced dense urban areas, freeways, tunnels, residential neighborhoods, airports, construction zones and pedestrian activity. That should give research access to the kinds of diverse scenarios self-driving vehicles will find themselves in, Ford says.

Plus, while most datasets only offer data from a single vehicle, this data comes from multiple vehicles. That means researchers can explore, for instance, what happens when two vehicles encounter each other. One might be able to detect things that the other cannot “see,” which could lead to development around multi-vehicle communication, perception and path planning.

The data is available through a collaboration with Amazon’s open data program, and more info can be found at avdata.ford.com. The first set of data is available now, and Ford says it will continue uploading the rest of the data logs.

Ford has tested its self-driving vehicles and delivery services in a handful of cities — though it clarified that this data is separate from the work it’s doing with Argo AI to develop a production-ready system. Ford still plans to launch its self-driving car fleet in 2021, but last spring, CEO Jim Hackett scaled back those ambitions. He said the fleet’s “applications will be narrow, what we call geo-fenced, because the problem is so complex.”