For Apple iCar? Forget about self-driving. Tim Cook is too smart to take the chance when there is a much cheaper and safer option...
Do you have an iPhone? Ever tried to unlock it using Face ID while wearing any kind of face covering? It doesn’t work, because the software is too brittle for a real-world application. Apple declares that Face ID is designed to work with hats, scarves, glasses, contact lenses and many sunglasses, but didn’t train its algorithms to work with face masks, let alone full-face respirators. Why would it?
In consumer electronics, brittle software has been a fact of life for decades. If you are over forty you will have experienced the horrors of the Blue Screen of Death. If you are younger and have never heard that name before, this is what it looks like.
In personal computing, brittle software means lost work; in cellphones it means dropped calls; in games consoles it means controllers hurled against the wall in frustration just as you finally defeated the boss. But in mission-critical environments such as aerospace, defense, industrial, medical and automotive, brittle software means a body count.
If Face ID won’t work on your iPhone because you are wearing a face covering, you just enter your passcode instead. No stress. However, if your car experiences fatal exception 0D at 0246:014747FB while traveling at highway speed, your final destination might turn out to be the morgue.
Broadly speaking, this is why cloud, computing and consumer electronics tech giants such as Apple, Google, Intel, Microsoft and Nvidia have made almost no progress in mission-critical automotive systems for the last several decades. So called “OEM automotive grade” development is hard and takes forever to validate and verify.
Uber seemingly didn’t think, or didn’t care, about the risks of hurling 4,500-pound Volvo XC90s around the streets of Tempe, Arizona, using test-level software and poorly trained safety drivers. It foolishly pursued the Silicon Valley mantra of “move fast and break things” in its efforts to be first to market with a self-driving robotaxi service and succeeded in killing Elaine Herzberg in March 2018.
Uber’s Advanced Technology Group (ATG) was sold to Aurora Innovation at the beginning of December, marking the end of its self-driving ambitions after having humiliated itself, incinerated capital and without ever saving a single life – and let’s not forget about abandoning the unfortunate safety driver (Rafaela Vazquez) to face charges of negligent homicide. History teaches us that, sometimes, the only difference between genius and stupidity is hindsight.
For the last five or so years, we have been watching a multi-billion-dollar game of poker play out under the guise of “self-driving.” Uber was the robotaxi joker and Tesla the sucker that showed everyone the liability risks associated with poorly designed automated driving systems and inadequate driver monitoring in mass-market vehicles.
Apple, in comparison, is neither stupid nor a sucker. Apple did what Apple does and sat on the sidelines and watched developments unfold. It waited. And waited. And waited.
Forget about a self-driving iCar
There is no shortage of news about “self-driving” developments and Waymo is undoubtedly making steady progress. However, glance through this video at the fabulously wide, mostly empty, beautifully sunlit streets of Arizona and decide for yourself how representative it is of your experience of everyday driving.
Self-driving is no longer the short-term goal for most mass-market automakers, who have come to realize that the system costs and liability risks are simply too great. Even SAE International is facing up to the reality that self-driving is a complex and cost-prohibitive feature in privately-owned passenger vehicles. In a recent article it concluded:
Enhanced SAE Level 2 automated driving systems avoid the cost, complexity – and uncertainty – of Level 3.
Readers of EE Times understood that 18 months ago, “For Mass-Market Cars, Forget L3-L5 Autonomy” and while the “J3016 Level Cops” will be all over me again on Twitter for saying this, the industry trend is clearly towards “Level 2+” where the human driver is liable at all times.
Apple is undoubtedly a brilliant consumer electronics company, but the iCar will be its first foray into the world of mission-critical, functionally-safe automotive system development. Rumors of the iCar sent lidar stocks soaring, but I’m left wondering why Apple would use lidar at all when it is most likely to use tested, proven and validated technology such as cameras and radar.
Apple will sell every last iCar that rolls off the production line and it doesn’t need to compete with Elon Musk and Tesla on anything. So, forget about a self-driving iCar, Tim Cook is too smart to take the chance when there is a much cheaper and safer option that doesn’t involve any legal danger or liability risk.
If the rumors that iCar will enter production in 2024 are accurate, then on automotive timescales the vehicle design is already mostly complete, even if only virtually as a “digital twin.” Apple will already be working with tier one and tier two suppliers, so we won’t have to look very far for some major clues.
My top candidate for tier one for the automated driving system is Veoneer, which has developed a solution that is different from almost everyone else, called “Collaborative Driving”.
Collaborative driving blends all the best features of human drivers and machine intelligence for ADAS and driver monitoring into a sophisticated driving system, but where the human driver is liable at all times. High performance driver state analysis will be used to monitor levels of attention, distraction, drowsiness, engagement and impairment, with warning and intervention strategies developed using human factors and behavioral research to keep the human driver in the loop.
Although Apple will need to work with a wide range of automotive suppliers to bring the iCar to production, it is evident that it has extensive proprietary knowledge in power management, battery technology, software-defined architectures and over-the-air updates. Apple also has significant relevant expertise in chip design (smartphone applications processors), operating system design (macOS and iOS) and human/machine interface design.
So, where I believe Apple will stun the automotive industry with the iCar is in infotainment, with an operating system I will name iCarOS, a series of car infotainment processors the first of which I will call the C1 (Car1) and robust, state-of-the-art eye-gaze tracking leading to an immersive in-cabin user experience the likes of which we can hardly begin to imagine.
Apple Face ID doesn’t work with face masks, so in a Covid-19 world the iCar will need a driver monitoring system that can handle everyday driving. Want to see what that looks like? Take a look:
In the next column I will share my thoughts on why I believe Apple will have gone “Down Under” for the driver monitoring system in the iCar and is likely to already be working with the clear technology leader, Seeing Machines. Until then, may I wish you a very happy and prosperous new year.