Nexar’s crowd-sourced vision network creates a massive dataset for AV training.  Each and every month, Nexar drivers “see” 130 million miles, generating tens of thousands of edge cases, collisions and near collisions. This data reflects different weather conditions, shows driving through construction zones, dangerous lanes and more, covering the long tail of corner cases, collisions and near collisions, all with GPS and accelerometer data. Additionally, Nexar can provide collision and near-collision rate statistics by geography and time, impact speed, weather conditions, G-force, actor types, impact geometry (hit direction) and more. Nexar’s data also makes use of its collision reconstruction technology that can determine the actions of the different objects and actors in a scene. For more information, go to data.getnexar.com

data.getnexar.com