Automotive electronics, with the Holy Grail being fully autonomous vehicles, is currently being touted as one of the biggest growth drivers for the semiconductor industry. So much so, that every event I’ve attended so far this year has featured sessions, presentations, keynotes, and panel discussions espousing the benefits of autonomous vehicles, the technologies that have been developed to enable them, the successful test runs, the remaining technical challenges (few), and even the costs (high). There is a distinct air of enthusiasm from those who spend their days researching and developing these technologies. Understandably so, because as engineers and technologists, inventing cars that drive themselves is exciting! To hear them speak about it, no challenge is insurmountable. In their eyes, the benefits of autonomous vehicles far outweigh the negatives. Once fully autonomous and driverless vehicles take over the roads — predicted to begin as soon as 2019 (for Tesla drivers), and become the norm sometime between 2025 and 2030 for the rest of us — they will make the world a better place. Or will they? To figure that out, we need to understand the various classifications of autonomous vehicles and the pros and cons of full autonomy.
Stages of Autonomy
Automotive autonomy is categorized into five levels. Level 1 means certain functions like automatic braking, stability control, and cruise control can be implemented, but only one at a time. Level 2 allows for multiple automated functions to work simultaneously. Essentially, Levels 1 and 2 are categorized as advanced driver assists systems (ADAS) that leave the driver in control, except in certain situations where safety is a concern. Levels 3-5 ramp up to full autonomy. MobileEye provides a detailed breakdown of the five levels here.
During ECTC 2017’s panel session Advanced Packaging for Autonomous Automobiles, panelists Venky Sundaram, Georgia Institute of Technology and Dongji Xie, Nvidia, both commented that the move to autonomy is already happening at the early levels and that full autonomy isn’t far behind.
Xie predicted that by 2020, there would already be 10M autonomous vehicles on the road and that auto-cruise and co-pilot functions are only the first step. Sundaram said that many OEMs have pledged to include Level 1 autonomous driving in all cars sold by 2022 to reduce the number car accident fatalities, as currently, 1.3M deaths globally are due to human error. Additionally, he said already in 2017, many automakers offer level 2 in their luxury models, and several offer level 2 as standard features.
Luxury vehicles like Tesla, Mercedes and others have already taken to the streets with level 3 and 4 autonomous test vehicles, proving technology readiness. However, with a price tag of $100K in 2020-2022, as predicted by panelist Frank Bertini of Velodyne, it could be quite a bit longer than predicted before these cars replace the family vehicle. In this recent interview with Ola Källenius, head of R&D at Mercedes, noted that the company is only developing Level 4 and 5 autonomy for robo-taxis because the systems cost “tens of thousands of dollars.” Although, according to this article on automakers timelines posted on VentureBeat by Dan Fagella, TechEmergence, premium Volvos that feature an autopilot system will cost an additional $10,000.
The Driverless Debate
ECTC panelist Brent Richardson, Texas Instruments, said the issues that could prevent autonomous vehicle technology from becoming mainstream will not be due to technical capability, but because of liability and infrastructure issues, as both human-piloted and driverless vehicles will likely have to share the roads for many years before human-piloted vehicles are phased out. In his list of points to ponder, Richardson asked, “what’s scarier; autonomous vehicles driving about, or teenagers taking the wheel?” In my opinion, it’s the answer to this question, and many more questions like them, that will ultimately determine the real future of fully autonomous vehicles.
Kathy Winter Automated Driving Group General Manager, Intel, has made a cross-country trip in a fully autonomous vehicle and presented a fairly convincing argument during her keynote at SEMICON West. She evangelized about the benefits of a world driven (quite literally) by autonomous vehicles vs. humans: It’s safer. The maps and logistics are better. It will lead to more vehicle sharing and therefore fewer cars on the road. She predicted $1.3T savings to the US economy and 500B accident cost reduction. Automated fleets will alleviate the shortage of drivers. It will enable new services that people in the vehicle could be consuming if they weren’t so busy driving. Sure, there are still issues to be ironed out – like rolling out 5G to support daily data usage of 4TB per day, and security issues that Winter says will be solved by establishing industry-wide standards and sharing of safety data. But these solutions are in the works. Winter urged us all to “let go of the wheel and embrace a passenger economy,” which she predicted will generate $7T in revenue from mobility and pilotless vehicle services (Figure 2).
The Missing Link
There’s no argument that ADAS features like lane departure warnings, park assist, rear traffic alert, and the ability for the car to take over control from distracted or drowsy drivers have distinct safety advantages. And while theoretically, Winter and other proponents of this passenger economy could be spot on with their long list of advantages, they are leaving out a significant factor that could delay or prevent the adoption of fully autonomous vehicles altogether: consumer buy-in.
When questioned about their opinions as consumers, even the very people who are developing these technologies aren’t so sure they’re ready to make that switch. How do I know this? I’ve been asking them. They all like to drive. The panelists at ECTC said ideally they’d own two cars; a fully autonomous one to commute downtown, and a human-piloted car for pleasure drives. (I’m pretty sure they were joking because this would be expensive.)
The two-car solution may be moot anyway, in the eventuality that human driving is outlawed like Elon Musk suggested to the Washington Post in 2015. In fact, according to this article on CNBC, the biggest problem faced by robo-taxis is having to share the road with human-piloted vehicles. One alternative to outlawing human driving would be to design and build a highway and road system that separates driverless and human-piloted vehicles. But that will take enormous resources and decades to achieve, and does nothing to achieve the goal of fewer cars on the road.
In my opinion, it boils down to this: before we can rely on driverless vehicles to solve both safety issues and the world’s transportation problems, consumers will have to not only give up the right to drive but be willing to rely on ride sharing services and driverless public transportation systems as well. The passenger economy will need to be fully embraced as a way of life for everything from trips to the grocery store, chauffeuring the kids, and commuting, to destination-less drives and family road trips. And while there are definite advantages to that scenario, convincing society as a whole that they’ll be better off handing over control of the wheel is going to be a tough sell. ~ FvT