Leverage Nexar’s collision and near-collision data, driver behavioral maps, change detection, mapping and safety data for AV success
The world as it is, ready for training
Nexar collects data from 160 Million Miles driven monthly, adding to a dataset of 25 million corner case videos. This imagery translates into valuable information for AV training, using various driving events such as hard brakes, collisions, near-misses, and other edge cases - all for improving your model’s performance and safety.
Accurate and fresh map data for AVs
Nexar’s dash cam network maps transient road elements such as work zones, detects changes to maps, from turn signs to speed limits, and helps you fill in the accurate map data you need, at scale and on time, with the visual verification you need.
Scenario reconstruction to benchmark incidents
Nexar produces corner case scenarios for simulation, using 3D reconstruction on top of collision and near-collision data. This data includes full 1st party and 3rd party reconstruction, as well as raw data (such as geo-location and timestamps) and additional insights for training outcomes.
Driver Behavioral Maps
Driver behavioral maps are based on crowd-sourced driving insights for different road segments, driver types, weather and other road conditions. They capture the information about actual human driving, and overlay it on a map, for autonomous and assisted driving. This data is overlaid on the base map as a high definition map layer showing human behavior on the maps that AVs use for driving. AVs can use it to know when to switch lanes before a turn, how to decelerate when cornering, where virtual stop lines are and more.
Nexar dash cams “see” them all
- From near-misses and low-impact collisions (fender bender) to high impact collisions (car crashes) in a given city
- Valuable visual data edge cases, including positional information of drivers, hard brakes, lane drift etc.
- Real-world changes that need to be mapped for AVs, such as construction zones, road sign changes, road blockages and more.
- Diverse data across time (days, weeks or months) and driver types (consumer, commercial), including in difficult weather, geo-location and additional variables.