Video Imaging Stack

DriveOS enables a full set of features to build an Autonomous Vehicle (AV) compute pipeline, which includes:

• Receiving high-resolution camera images, RADAR, and Lidar from sensor modules.

• Processing images for computer and human vision.

• Enabling image pre-processing such as Lense Distortion Correction, scaling, cropping, etc.

The processed sensor inputs are used by Deep Learning algorithms to output actual commands. In parallel, sensor output can be encoded and saved to storage and rendered to user for vehicle surround visualization, rear, or side view.

The block diagram below describes the data and control flow between different engines and APIs that program the engine.