Friday, Feb 11, 2022, 15:50 Others

Will The Apple Car Drive Itself? – Autonomous Driving Thanks To AI & Machine Learning

Adaptive cruise control, active lane and course detection, cross traffic alerts, and pedestrian recognition with emergency braking assistance are all features that – along with a few other helpers, have become standard features of at least mid-upper tier automobiles. At times, the combination of features even appears so seamless that a driver might be convinced that their vehicle could drive itself. However, this is more illusion than reality – as the complexity of modern day traffic often necessitates very quick reaction times still unattainable without human intervention. Thus, the reason that these features are called "assistive" features and not entirely autonomous features. As it currently stands, even the most advanced machines and software are still incapable of outperforming advanced machines and software supervised by humans.

Assistive Systems Still Aren't Perfect
The perfection of assistive driving systems is currently the biggest challenge that the industry faces in the development of fully autonomous vehicles. The employees at Cupertino responsible "Project Titan" are currently hard at work on developing a solution to the problem for the upcoming Apple Car. The proof is a patent filed by Apple over 3 years ago just published by the US Patent & Trademark Office. The application carries the title "Evaluating varying-sized action spaces using reinforcement learning". More than anything else, Apple is working on the vehicle's decision mechanisms. The Californian company wants to considerably improve this feature in particular in comparison to current options on the market.



Source: Apple

Apple Car To Master Machine Learning
According to Apple, there have been significant improvements to hardware and software over the last few years – however, they still aren't sufficient enough to satisfy future automotive requirements. The patent application states that there's still a great deal of improvement to be made when it comes to evaluating the environment. Assistive systems for autonomous vehicles need to be able to work considerably faster than they do to prevent impending collisions. Apple relies on artificial intelligence and machine learning (AI and ML) for help, using a combination of data not only from the vehicle equipped with the system, but also from information and "experiences" collected from other vehicles. AI and ML then process possible reactions from the information and experiences for all conceivablly simulable scenarios to take part in the next decision process (and so on and so forth).

Quick Reaction To Critical Situations
Apple describes a set of situation in the patent application which would necessitate an AI/ML equipped vehicle. These would then be able to determine whether the best course of action is to turn out of the way or simply to change course. This should, of course, eventually include appropriate reactions to unforeseeable maneuvers from other drivers. Whether or not Apple has already begun developing such a system as described is still unknown. However, it is very easy to tell how Apple plans to accomplish its Apple Car ambitions – "Project Titan" is being led by John Giannandrea, the company's Senior Vice President of Machine Learning and AI Strategy.

More articles you might enjoy to read: