Self-driving vehicles have long been touted as a means to drastically lower accidents and improve transportation efficiency.
However, analysts believe the arrival of autonomous driving tech will be slowed, probably one to two years, because of the recent drastic downturn in auto sales and production due to COVID-19. Depending on whether the virus reappears a second time, the slowdown could last longer.
FierceElectronics interviewed five analysts who have close contact with chipmakers and auto makers working on revolutionary autonomous vehicle (AV) technology. Most said that chip designers will slow down their R&D efforts in coming quarters, even as semi makers and car companies continue to believe in the value of the technology over the long term.
Some carmakers, like China’s Xpeng Motors using Nvidia chips, are even moving ahead with new driving automation, although it isn’t clear how many in the industry will follow suit.
“One of the effects of COVID-19 is that the semiconductor guys are expecting some delays of their [self-driving] development programs,” said Phil Amsrud, an analyst at IHS Markit, in an interview.
“It’s not so much a change in their priorities, but more are saying they are making less money this year, the pot is smaller and their investment in research will be smaller,” he said. “I think COVID will cause a bit of a pause as people look at the economic realities.” To be clear, no semi maker is saying any of their driving automation development work is dead or shelved, he added.
The current economic realities are about as harsh as one can imagine, with major car companies like GM and Ford closing plants and making ventilators instead. IHS Markit expects global vehicle sales to decline 22% this year to 70 million units, with a 26.6% fall in the U.S. to 12.5 million units, compared to 2019.
Survey finds 65% of IC makers see auto tech delays
A recent, unpublished IHS Markit survey of 46 companies in the auto integrated circuit (IC) supply chain found that nearly two-thirds are expecting a delay in technology deployments in upcoming launches. “Almost 65% said they see delays in technology deployment of upcoming projects,” Amsrud said. Just over 20% said there would be no delay of new technology or product development.
The companies in the survey are all semi suppliers who make a breadth of products including components, diodes, transistors, memory, microcontrollers and ASICs. More than 100 companies globally make IC products for ADAS (advanced driver assistance systems) and autonomous driving, according to IHS Markit.
Level 5 automation is like going to mars
Even before COVID-19 hit, there was a profound shift in the industry over the past two years to a focus on ADAS with its lower technology aspirations and an earlier return on results than on AV. The highest level of autonomous driving, Level 5, calls for complete automation where a vehicle has no steering wheel or pedals. “At that highest level, companies are now saying, ‘this is really hard and a lot harder than we thought it would be,’” Amsrud said.
Level 5 is now widely viewed as the vehicle automation equivalent of a mars shot, he added.
But even lower levels of automation will be affected by delays in R&D, including Level 3, where the vehicle itself controls all monitoring of the driving environment. through sensors like LiDAR. At that level, a vehicle that can’t detect roadway conditions with snow or fog would likely turn over all driving control to a human operator. Technology delays could include where a company was planning to introduce a new feature in 2021, but will delay deployment in a vehicle system for another model year. Or, it could be a semiconductor that was planned for roll out as a 7nm device will be held off longer to get more life out of an existing 14nm device, Amsrud said.
GSA survey finds virus has a negative impact on AV
A separate survey of 22 top-level executives at global semiconductor makers found 27% said COVID-19 would have a negative impact on investment, growth and adoption of autonomous vehicles. Another 9% said the virus would have a positive impact on AVs, while 59% said the impact would be neutral. The survey, conducted recently by KPMG and the Global Semiconductor Alliance, also found less negative impact of the virus on 5G, artificial intelligence and Internet of Things applications than on AV.
Long-term growth for ADAS/AV remains
IDC analyst Matt Arcaro said his company is planning to report in late May on what level of autonomy newly produced light-vehicles will have for each of the coming model years to 2024.
“Without vehicle manufacturing now, there will definitely be a a significant 2020 in-year impact to chipmakers,” he said. “I do think there likely be some slight residual impact in 2021. Based on what we can see now, I would expect that the long-term trajectory and growth story will remain consistent with what was expected pre-COVID-19.”
Arcaro was careful to say that lower chip market demand and revenue doesn’t necessarily mean less investment in R&D for auto chipsets. “Automotive, because of its scale and volumes, will remain a key area for growth for chipset providers,” he said. “We are still very early days in the automotive growth cycle for high compute autonomy platforms.”
Ironically, COVID-19 might even speed up the market forces that compel chip providers to focus their auto strategy more narrowly, Arcaro added. “Rather than focusing on a whole suite of AV chips, they may focus on just a subset of autonomy or look to form a business partnership or joint venture to de-risk some of the uncertain timelines and volumes for full driverless autonomy.”
Robo-taxis means no driver to spread germs, but development will slow
Some experts are suggesting that robot taxis will be more desirable to limit the spread of infectious diseases like COVID-19 (with no driver present to spread germs) in the future, but it isn’t clear that goal will quicken the pace of development.
“Robot taxis have long been put forward as the solution most appropriate for autonomy,” said Gartner analyst Michael Ramsey. “Those efforts won’t disappear, but they certainly will slow. Full autonomy will be pushed into more business-facing uses like mining, logistics, warehouse operations and other places where there is a good mix of needs and the potential for efficiency and cost savings.”
Ramsey said the longer trend is that the autonomous market will become more diverse and not driven by the large scale of the auto industry, at least for really advanced autonomy. High-tech chips in vehicles will be focused heavily on ADAS, he predicted.
No announced postponements so far
Gaurav Gupta, another Gartner analyst, said he hasn’t heard from leading auto chipmakers about postponing any of their launches so far. “Automakers are burning through cash now as vehicle sales have nosedived and as a result one would assume their R&D spending would be conservative in the near future and would focus on technology that can help them with revenue in the short term,” he said.
Some companies like Mobileye, a division of Intel, have reported higher revenues in the latest quarter with lower revenues expected over the rest of the year due to lower vehicle sales. But that doesn’t include their estimate for R&D spend, which is the true for other chipmakers reporting recent earnings.
“Big leaders will continue their push for R&D and investment in the ADAS and AV space and perhaps look for acquisitions,” Gupta added. Meanwhile, the ecosystem for companies researching AV products is due for consolidation with many small players, he and other analysts said.
China’s Xpeng and the brand new P7
One of latest innovations in Level 3 automated driving came out recently with Nvidia Drive AGX Xavier AI technology in the Xpeng P7, an all-electric sports sedan introduced this week in China. It relies on continuous over-the-iar software updating to fine tune the AI.
Nvidia first described Xavier at CES in 2018 to enable Level 3 features that allow steering, acceleration/deceleration and passing of other cars without human input. Level 3 allows drivers to take their hands off the wheel and feet off the pedals but humans still need to be ready to take back control when the car requests it. The Audi A8 was expected to have level 3 capabilities, but the company recently announced it will hold off adding in the Traffic Jam Pilot system. **
The P7 achieves Level 3 with 12 ultrasonic sensors, five millimeter-wave radars and 14 cameras that offer 360-degree perception. Notably, there’s no LiDAR. The forward-looking radar can detect obstacles at 200 meters, penetrating rain, fog and haze, Nvidia said. The P7’s starting price is about $34,000.
At CES 2020 in January, Qualcomm launched its Snapdragon Ride automated platform for assisted and self-driving, offering FierceElectronics a 20-minute test drive along interstates at speeds more than 60 mph. A safety driver was seated behind the wheel but he didn’t steer or put his feet on the brakes or accelerator for the highway porition of the demo of the Qualcomm tech in a Lincoln MKZ.
While automated driving continues to appear in some cars, it isn’t clear how much the technology will be delayed or the research cut back beyond mid-2021. Much depends on whether COVID-19’s spread gets worse or reappears and further crushes the car industry.
“AV and ADAS are really more of a long-term play,” said Jack Gold, an analyst at J. Gold Associates with a more optimistic outlook. “Companies may have to reduce some expenditures if they are badly affected by the market downturn and their revenues tank, but I for the most part I don’t see any significant pulling back for R&D for future deployments of AV and ADAS.”
The reality of COVID-19 is hitting all sorts of tech-centric industries, especially semiconductors, where the lead times for researching and developing products can stretch out for years. “If you don’t invest now, you won’t have product when it is needed in the next couple of years,” Gold said.
** Audi's decision to not add Traffic Jam Pilot to the A8 was not contained in the original version of this story.