Apple has officially released its self-driving car research, marking the first time the company has publicly disclosed an autonomous vehicle paper.
The publication of the paper — which was written by Yin Zhou and Oncel Tuzel and was submitted on November 17th — is notable because Apple has traditionally been secretive when it comes ot its upcoming products, particularly those pertaining to artificial intelligence and machine learning.
In their report, the scientists propose a new software approach called “VoxelNet,” which is intended to help computers better detect three-dimensional objects.
Typically, self-driving cars adopt normal two-dimensional cameras and depth-sensing “LiDAR” units to scan surrounding environments. However, the cameras’ low resolutions make it difficult to detect small, distant objects.
Using VoxelNet, though, Apple researchers said they were able to get “highly encouraging results” related to spotting pedestrians and cyclists with just LiDAR data. Additionally, they said they were able top other approaches for detecting 3D objects that use only LiDAR. The researchers said noted that tests were conducted in computer simulations, not on the road.
Apple CEO Tim Cook has previously called self-driving cars “the mother of all AI projects.” The company was granted a permit in California to test autonomous vehicles back in April.
Source: Reuters
