Apple publishes research on 3D object detection for autonomous cars

Tim Cook has already warned us: yes, Apple is working on a system for autonomous cars.

For now, as you can imagine, Ma is still in the research process; but apparently, this time she wants to share her findings or part of them with the rest of the world. This is because, according to ReutersApple scientists have published research (PDF) on how autonomous cars can better identify cyclists and pedestrians while using fewer sensors.

Published November 17, the paper basically explains how accurate detection of 3D point cloud objects can be used for autonomous navigation, cleaning robots, and augmented or virtual reality.

Apple Research on Autonomous Car System

The authors of the article, Yin Zhou and Oncel Tuzel, who have been working on machine learning research at Apple for two years, suggested a new software approach called "VoxelNet" that would help computers detect three-dimensional objects.

In this paper, we removed the need for manual resource engineering for 3D point clouds and proposed VoxelNet, a generic 3D detection network that unifies resource extraction and bounding box prediction into a single end-to-end exchange network.

Scientists seem certain that their research shows a very optimistic future for detecting 3D objects to identify automobiles, cyclists and pedestrians, claiming that their technology achieves better performance than they already exist today.

The full search can be accessed at this link (PDF).

Looking at the pictures, I think with my buttons: So, was this what this Lexus RX450h was doing on the streets of California?

via 9to5Mac