System and method of using multi-resolution camera clusters input data for controlling autonomous vehicles

Abstract

In one example, a method may include capturing two-dimensional (2D) images of scenes using a set of multi-resolution cameras disposed on at least one side of an autonomous vehicle. Further, a low-resolution depth map with relatively small depths may be generated for each scene using the captured 2D images. Furthermore, a high-resolution depth map may be generated for each scene for a wide depth range by iteratively refining the low-resolution depth map for each scene. Also, a 3D video may be generated based on the high-resolution depth maps and the captured 2D images of the central camera. Further, a distance, a velocity, and/or an acceleration of one or more objects relative to the autonomous vehicle is computed by analyzing one or more frames of the 3D video. Then, the autonomous vehicle may be controlled based on the computed distance, velocity, and/or acceleration of the one or more objects.