In a recently published research paper, Apple computer scientists revealed a new method they had developed, which could help fine-tune self-driving technology. The method could help machines detect any pedestrians and cyclists in its path.
The paper, which was published on November 17, offers a rare glimpse into Apple’s forays into the world of research technology.
Authored by Apple researchers Yin Zhou and Oncel Tuzel, the paper discusses advancements the researchers have made in the field of 3-D object detection. The paper was published in an international web journal known as arXiv.
The paper states that current object detection systems all rely on “LiDAR” technology. This technology works on the principle of speed calculation. A light is used to illuminate the surface and the speed at which the light returns is used to calculate its distance. This in turn is used to create an image of the object’s shape.
While LiDAR functions well with larger objects, it makes the detection of smaller objects slightly troublesome.
Additionally, LiDAR works best for objects that are at smaller distances from the machine. In order to assess an object located farther away, the technology would require support from an additional camera.
Apple’s researchers also argue that the entire setup could be “more sensitive to sensor failure”.
Thus, instead of the additional camera, Apple’s scientists have proposed a detection method based solely on LiDAR. They have christened the method “VoxelNet”.
VoxelNet applies complex computer vision and AI technology to carry out its functions.
Although Apple has only tested its new technology in computer simulations, it claims that VoxelNet “outperforms the state-of-the-art LiDAR-based-3-D detection methods by a large margin”.
Apple’s advent into self- driving vehicles has been kept under wraps for a long time. Apart from a few tidbits offered by Apple CEO Tim Cook, the company has remained tight lipped on the issue.
However, sources state that Apple was granted a permit to test self-driving cars by Californian authorities earlier this year.