Toggle Dark Mode
Apple has been extremely quiet about its self-driving car project, but the veil of secrecy might slowly be lifting. On Friday, Apple researchers gave a rare presentation of some of the machine-learning tech they’re developing.
Ruslan Salakhutdinov, Apple’s AI director, discussed several of the company’s machine-learning projects related to self-driving vehicles at the Neural Information Processing Systems conference in Long Beach, California. Reportedly, Apple’s presentations were meant to show off the technology it’s been working on — and to attract potential recruits, according to the MIT Technology Review.
One of Apple’s projects, which Salakhutdinov showed off at the conference, was a method for training systems to recognize pedestrians and other vehicles from 3D point cloud information. Interestingly, Apple first disclosed this technology in a research paper it had published online last month.
But Cupertino revealed other previously undisclosed technology at the conference, too. One project was a system for identifying and classifying different objects using cameras mounted on a vehicle. Another system used similar camera footage to precisely pinpoint and track a self-driving vehicle’s position.
This technique, as Technology Review points out, is known as simultaneous location and mapping, or SLAM. And while it can be used for self-driving vehicles, it also has potential uses in the augmented and virtual reality spheres.
Apple’s self-driving car project has been a closely held secret for years (the company has never publicly acknowledged the tech). But details about the so-called Project Titan have still leaked, and Apple CEO Tim Cook even confirmed the rumors in an interview with Bloomberg earlier this year, stating that the company is “focused on autonomous systems,” which he calls the “mother of all AI projects.”
During Apple’s presentations, another AI researcher, Carlos Guestrin, spoke about other machine-learning projects at Cupertino. Guestrin is Apple’s Seattle-based director of machine learning, and joined the company after Apple acquired his startup, Turi, last year. Those other projects included the advanced facial recognition system that powers Face ID, as well as a new tool called Turi Create that’s designed to make machine-learning easier to perform.
Apple is one of the foremost researchers in machine-learning and AI technology, but the company has been famously secretive about its developments in the field. Until fairly recently, Apple’s researchers were barred from publishing their work so as not to tip off its competitors about future products or services.
Despite that, Guestrin reassured his colleagues and potential recruits that working at Apple does not mean that their research work won’t be published. As evidence of that, Apple launched a new blog called Machine Learning Journal, which it has expanded on and detailed several of its AI-related technologies and projects.
Still, the company only reveals a small amount of its research when compared to other AI giants like Google and Facebook. So it’s likely that Apple is working on some interesting things, unbeknownst to the public.